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": 10.691908264160157,
|
|
"epoch": 0.000292243848266994,
|
|
"grad_norm": 12.625,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.8406,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 11114.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691956710815429,
|
|
"epoch": 0.000584487696533988,
|
|
"grad_norm": 13.375,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7353,
|
|
"mean_token_accuracy": 0.00011737089371308685,
|
|
"num_tokens": 20935.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.690714645385743,
|
|
"epoch": 0.000876731544800982,
|
|
"grad_norm": 9.4375,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.5144,
|
|
"mean_token_accuracy": 0.03050256703281775,
|
|
"num_tokens": 31930.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.654311275482177,
|
|
"epoch": 0.001168975393067976,
|
|
"grad_norm": 5.96875,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.1606,
|
|
"mean_token_accuracy": 0.05743609480559826,
|
|
"num_tokens": 41672.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.475521564483643,
|
|
"epoch": 0.00146121924133497,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.9607,
|
|
"mean_token_accuracy": 0.05290710777044296,
|
|
"num_tokens": 52510.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.543301010131836,
|
|
"epoch": 0.001753463089601964,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.8154,
|
|
"mean_token_accuracy": 0.045297696441411975,
|
|
"num_tokens": 63214.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.49622220993042,
|
|
"epoch": 0.002045706937868958,
|
|
"grad_norm": 2.5,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.7008,
|
|
"mean_token_accuracy": 0.05302377603948116,
|
|
"num_tokens": 73942.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.475418663024902,
|
|
"epoch": 0.002337950786135952,
|
|
"grad_norm": 2.375,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.7051,
|
|
"mean_token_accuracy": 0.05076396390795708,
|
|
"num_tokens": 84526.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.52062005996704,
|
|
"epoch": 0.0026301946344029457,
|
|
"grad_norm": 2.234375,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.7003,
|
|
"mean_token_accuracy": 0.04803669825196266,
|
|
"num_tokens": 95178.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.508524227142335,
|
|
"epoch": 0.00292243848266994,
|
|
"grad_norm": 2.3125,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.6047,
|
|
"mean_token_accuracy": 0.05312369242310524,
|
|
"num_tokens": 106653.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.445098209381104,
|
|
"epoch": 0.0032146823309369336,
|
|
"grad_norm": 2.25,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.4922,
|
|
"mean_token_accuracy": 0.058677663654088975,
|
|
"num_tokens": 115969.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.33490447998047,
|
|
"epoch": 0.003506926179203928,
|
|
"grad_norm": 2.09375,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.469,
|
|
"mean_token_accuracy": 0.06182386502623558,
|
|
"num_tokens": 125826.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.331936168670655,
|
|
"epoch": 0.0037991700274709215,
|
|
"grad_norm": 2.3125,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.4398,
|
|
"mean_token_accuracy": 0.06020851917564869,
|
|
"num_tokens": 136677.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.381442165374756,
|
|
"epoch": 0.004091413875737916,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.3696,
|
|
"mean_token_accuracy": 0.05836572460830212,
|
|
"num_tokens": 147535.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.336608695983887,
|
|
"epoch": 0.00438365772400491,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.2591,
|
|
"mean_token_accuracy": 0.06145334914326668,
|
|
"num_tokens": 157524.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.270106220245362,
|
|
"epoch": 0.004675901572271904,
|
|
"grad_norm": 1.6953125,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.1922,
|
|
"mean_token_accuracy": 0.063629350066185,
|
|
"num_tokens": 168344.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.35278034210205,
|
|
"epoch": 0.004968145420538897,
|
|
"grad_norm": 1.953125,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.14,
|
|
"mean_token_accuracy": 0.05551427491009235,
|
|
"num_tokens": 178976.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.23039789199829,
|
|
"epoch": 0.0052603892688058915,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.0429,
|
|
"mean_token_accuracy": 0.06621964909136295,
|
|
"num_tokens": 190567.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.196823787689208,
|
|
"epoch": 0.005552633117072886,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.9449,
|
|
"mean_token_accuracy": 0.07107578478753566,
|
|
"num_tokens": 201587.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.113740062713623,
|
|
"epoch": 0.00584487696533988,
|
|
"grad_norm": 1.6953125,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.8501,
|
|
"mean_token_accuracy": 0.06472622714936734,
|
|
"num_tokens": 211913.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.149368000030517,
|
|
"epoch": 0.006137120813606874,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.7762,
|
|
"mean_token_accuracy": 0.06379772424697876,
|
|
"num_tokens": 222247.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 9.989063167572022,
|
|
"epoch": 0.006429364661873867,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.6212,
|
|
"mean_token_accuracy": 0.07087173387408256,
|
|
"num_tokens": 232361.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 9.965903949737548,
|
|
"epoch": 0.006721608510140861,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.5651,
|
|
"mean_token_accuracy": 0.06980551891028881,
|
|
"num_tokens": 242885.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.803161907196046,
|
|
"epoch": 0.007013852358407856,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.4687,
|
|
"mean_token_accuracy": 0.06787880472838878,
|
|
"num_tokens": 254482.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.760642433166504,
|
|
"epoch": 0.00730609620667485,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.3483,
|
|
"mean_token_accuracy": 0.06577648222446442,
|
|
"num_tokens": 265744.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.514495658874512,
|
|
"epoch": 0.007598340054941843,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.2621,
|
|
"mean_token_accuracy": 0.0672954872250557,
|
|
"num_tokens": 276924.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.37683687210083,
|
|
"epoch": 0.007890583903208837,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.1773,
|
|
"mean_token_accuracy": 0.07101362608373166,
|
|
"num_tokens": 287779.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.18982000350952,
|
|
"epoch": 0.008182827751475831,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.1052,
|
|
"mean_token_accuracy": 0.07210481017827988,
|
|
"num_tokens": 298807.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.113053703308106,
|
|
"epoch": 0.008475071599742826,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.1007,
|
|
"mean_token_accuracy": 0.061491196230053904,
|
|
"num_tokens": 309760.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.824873638153075,
|
|
"epoch": 0.00876731544800982,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.9698,
|
|
"mean_token_accuracy": 0.0695835679769516,
|
|
"num_tokens": 319750.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.6828182220459,
|
|
"epoch": 0.009059559296276814,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.8822,
|
|
"mean_token_accuracy": 0.07002829425036908,
|
|
"num_tokens": 329931.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.598028564453125,
|
|
"epoch": 0.009351803144543808,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.9106,
|
|
"mean_token_accuracy": 0.06758572831749916,
|
|
"num_tokens": 340457.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.501685905456544,
|
|
"epoch": 0.009644046992810802,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.8105,
|
|
"mean_token_accuracy": 0.07629426941275597,
|
|
"num_tokens": 352450.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.418387317657471,
|
|
"epoch": 0.009936290841077795,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.8297,
|
|
"mean_token_accuracy": 0.07713140919804573,
|
|
"num_tokens": 363204.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.363510417938233,
|
|
"epoch": 0.010228534689344789,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.8784,
|
|
"mean_token_accuracy": 0.06442252397537232,
|
|
"num_tokens": 373577.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.313191223144532,
|
|
"epoch": 0.010520778537611783,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.7833,
|
|
"mean_token_accuracy": 0.07531154453754425,
|
|
"num_tokens": 384267.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.214359951019286,
|
|
"epoch": 0.010813022385878777,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.7491,
|
|
"mean_token_accuracy": 0.06811888329684734,
|
|
"num_tokens": 394767.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.22794075012207,
|
|
"epoch": 0.011105266234145771,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.7665,
|
|
"mean_token_accuracy": 0.07282254844903946,
|
|
"num_tokens": 405807.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.217875862121582,
|
|
"epoch": 0.011397510082412765,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.7191,
|
|
"mean_token_accuracy": 0.06757043041288853,
|
|
"num_tokens": 416381.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.152586936950684,
|
|
"epoch": 0.01168975393067976,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.8287,
|
|
"mean_token_accuracy": 0.07003305219113827,
|
|
"num_tokens": 427262.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.141828060150146,
|
|
"epoch": 0.011981997778946754,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.7651,
|
|
"mean_token_accuracy": 0.0717297799885273,
|
|
"num_tokens": 438504.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.196744251251221,
|
|
"epoch": 0.012274241627213748,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.7513,
|
|
"mean_token_accuracy": 0.07213160172104835,
|
|
"num_tokens": 449923.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.13697328567505,
|
|
"epoch": 0.01256648547548074,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.7046,
|
|
"mean_token_accuracy": 0.07478658258914947,
|
|
"num_tokens": 460736.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.1306658744812,
|
|
"epoch": 0.012858729323747735,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.8163,
|
|
"mean_token_accuracy": 0.06586915515363216,
|
|
"num_tokens": 471513.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.089940357208253,
|
|
"epoch": 0.013150973172014729,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.6608,
|
|
"mean_token_accuracy": 0.07503279075026512,
|
|
"num_tokens": 482115.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.06955795288086,
|
|
"epoch": 0.013443217020281723,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.6789,
|
|
"mean_token_accuracy": 0.07244161255657673,
|
|
"num_tokens": 493273.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 8.088439989089967,
|
|
"epoch": 0.013735460868548717,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.7067,
|
|
"mean_token_accuracy": 0.07398871332406998,
|
|
"num_tokens": 503939.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.001152181625367,
|
|
"epoch": 0.014027704716815711,
|
|
"grad_norm": 0.8671875,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.6139,
|
|
"mean_token_accuracy": 0.07592850960791112,
|
|
"num_tokens": 515215.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.98775544166565,
|
|
"epoch": 0.014319948565082705,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.7051,
|
|
"mean_token_accuracy": 0.07671321779489518,
|
|
"num_tokens": 526485.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.02779312133789,
|
|
"epoch": 0.0146121924133497,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.6616,
|
|
"mean_token_accuracy": 0.07241038978099823,
|
|
"num_tokens": 536735.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.066807174682618,
|
|
"epoch": 0.014904436261616694,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.7528,
|
|
"mean_token_accuracy": 0.06985807269811631,
|
|
"num_tokens": 547326.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.9721503257751465,
|
|
"epoch": 0.015196680109883686,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.6367,
|
|
"mean_token_accuracy": 0.07485130280256272,
|
|
"num_tokens": 558275.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.969989395141601,
|
|
"epoch": 0.01548892395815068,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.6104,
|
|
"mean_token_accuracy": 0.0731944501399994,
|
|
"num_tokens": 569711.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.970643997192383,
|
|
"epoch": 0.015781167806417674,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.5828,
|
|
"mean_token_accuracy": 0.07749090045690536,
|
|
"num_tokens": 579602.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.9638073444366455,
|
|
"epoch": 0.01607341165468467,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.6231,
|
|
"mean_token_accuracy": 0.07638676390051842,
|
|
"num_tokens": 590899.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.931737279891967,
|
|
"epoch": 0.016365655502951663,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.6258,
|
|
"mean_token_accuracy": 0.07663728073239326,
|
|
"num_tokens": 602080.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.949186515808106,
|
|
"epoch": 0.016657899351218655,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.6619,
|
|
"mean_token_accuracy": 0.07214237600564957,
|
|
"num_tokens": 613692.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.928168773651123,
|
|
"epoch": 0.01695014319948565,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.6255,
|
|
"mean_token_accuracy": 0.07801526971161366,
|
|
"num_tokens": 625019.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.858082962036133,
|
|
"epoch": 0.017242387047752643,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.5324,
|
|
"mean_token_accuracy": 0.07193354479968547,
|
|
"num_tokens": 636453.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.900658082962036,
|
|
"epoch": 0.01753463089601964,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.5805,
|
|
"mean_token_accuracy": 0.07368646003305912,
|
|
"num_tokens": 647272.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.980489492416382,
|
|
"epoch": 0.017826874744286632,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.5929,
|
|
"mean_token_accuracy": 0.07068270668387414,
|
|
"num_tokens": 657724.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.874118614196777,
|
|
"epoch": 0.018119118592553628,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.5913,
|
|
"mean_token_accuracy": 0.07631154656410218,
|
|
"num_tokens": 669183.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.852222537994384,
|
|
"epoch": 0.01841136244082062,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.5557,
|
|
"mean_token_accuracy": 0.07173253633081914,
|
|
"num_tokens": 680439.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.902084875106811,
|
|
"epoch": 0.018703606289087616,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.6115,
|
|
"mean_token_accuracy": 0.07198056317865849,
|
|
"num_tokens": 690789.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.909448766708374,
|
|
"epoch": 0.01899585013735461,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.5829,
|
|
"mean_token_accuracy": 0.07492739222943783,
|
|
"num_tokens": 701818.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.837027454376221,
|
|
"epoch": 0.019288093985621604,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.5366,
|
|
"mean_token_accuracy": 0.07418245449662209,
|
|
"num_tokens": 713344.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.862662744522095,
|
|
"epoch": 0.019580337833888597,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.5289,
|
|
"mean_token_accuracy": 0.07505915723741055,
|
|
"num_tokens": 723279.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.862085199356079,
|
|
"epoch": 0.01987258168215559,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.5211,
|
|
"mean_token_accuracy": 0.07510280013084411,
|
|
"num_tokens": 733875.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.86326003074646,
|
|
"epoch": 0.020164825530422585,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.5595,
|
|
"mean_token_accuracy": 0.073441531509161,
|
|
"num_tokens": 745449.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.847935771942138,
|
|
"epoch": 0.020457069378689578,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.5717,
|
|
"mean_token_accuracy": 0.07788249254226684,
|
|
"num_tokens": 757570.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.815750980377198,
|
|
"epoch": 0.020749313226956573,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.4513,
|
|
"mean_token_accuracy": 0.08608488291501999,
|
|
"num_tokens": 767601.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.804518222808838,
|
|
"epoch": 0.021041557075223566,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.5134,
|
|
"mean_token_accuracy": 0.07662104293704033,
|
|
"num_tokens": 777852.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.8234796047210695,
|
|
"epoch": 0.021333800923490562,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.4728,
|
|
"mean_token_accuracy": 0.07879027798771858,
|
|
"num_tokens": 788339.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.763150691986084,
|
|
"epoch": 0.021626044771757554,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.5509,
|
|
"mean_token_accuracy": 0.07561626769602299,
|
|
"num_tokens": 799791.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.7915912628173825,
|
|
"epoch": 0.02191828862002455,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.4559,
|
|
"mean_token_accuracy": 0.07913151681423188,
|
|
"num_tokens": 810141.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.82445068359375,
|
|
"epoch": 0.022210532468291543,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.4938,
|
|
"mean_token_accuracy": 0.07601816579699516,
|
|
"num_tokens": 820929.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.806576347351074,
|
|
"epoch": 0.022502776316558535,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.4415,
|
|
"mean_token_accuracy": 0.07921079695224761,
|
|
"num_tokens": 831340.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.650366687774659,
|
|
"epoch": 0.02279502016482553,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.4444,
|
|
"mean_token_accuracy": 0.07921848297119141,
|
|
"num_tokens": 841846.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.847888851165772,
|
|
"epoch": 0.023087264013092523,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.4322,
|
|
"mean_token_accuracy": 0.0797240749001503,
|
|
"num_tokens": 852089.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.713356971740723,
|
|
"epoch": 0.02337950786135952,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 7.4116,
|
|
"mean_token_accuracy": 0.08028053492307663,
|
|
"num_tokens": 862568.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.74327917098999,
|
|
"epoch": 0.02367175170962651,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000202,
|
|
"loss": 7.4684,
|
|
"mean_token_accuracy": 0.07664722427725792,
|
|
"num_tokens": 874676.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.674819469451904,
|
|
"epoch": 0.023963995557893508,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 7.3829,
|
|
"mean_token_accuracy": 0.08288070000708103,
|
|
"num_tokens": 886338.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.71968150138855,
|
|
"epoch": 0.0242562394061605,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 7.4114,
|
|
"mean_token_accuracy": 0.07737804017961025,
|
|
"num_tokens": 896671.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.6655017852783205,
|
|
"epoch": 0.024548483254427496,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.3914,
|
|
"mean_token_accuracy": 0.07870943620800971,
|
|
"num_tokens": 908265.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.6367028713226315,
|
|
"epoch": 0.02484072710269449,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.000212,
|
|
"loss": 7.3495,
|
|
"mean_token_accuracy": 0.0839125782251358,
|
|
"num_tokens": 919024.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.721334648132324,
|
|
"epoch": 0.02513297095096148,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 7.3314,
|
|
"mean_token_accuracy": 0.08483307361602783,
|
|
"num_tokens": 930837.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.614313220977783,
|
|
"epoch": 0.025425214799228477,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 7.3096,
|
|
"mean_token_accuracy": 0.08256931453943253,
|
|
"num_tokens": 941374.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.603088331222534,
|
|
"epoch": 0.02571745864749547,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 7.2577,
|
|
"mean_token_accuracy": 0.08504624664783478,
|
|
"num_tokens": 953273.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.580397319793701,
|
|
"epoch": 0.026009702495762465,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 0.000222,
|
|
"loss": 7.2375,
|
|
"mean_token_accuracy": 0.0916403166949749,
|
|
"num_tokens": 964882.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.614603757858276,
|
|
"epoch": 0.026301946344029457,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 7.4103,
|
|
"mean_token_accuracy": 0.08159535191953182,
|
|
"num_tokens": 974732.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.595666790008545,
|
|
"epoch": 0.026594190192296453,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 7.3392,
|
|
"mean_token_accuracy": 0.07731116712093353,
|
|
"num_tokens": 985709.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.649027681350708,
|
|
"epoch": 0.026886434040563446,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 7.3048,
|
|
"mean_token_accuracy": 0.08609704971313477,
|
|
"num_tokens": 998177.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.5427803039550785,
|
|
"epoch": 0.02717867788883044,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 7.2407,
|
|
"mean_token_accuracy": 0.08310350701212883,
|
|
"num_tokens": 1008676.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.5991309642791744,
|
|
"epoch": 0.027470921737097434,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 7.357,
|
|
"mean_token_accuracy": 0.08094775602221489,
|
|
"num_tokens": 1019951.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.504101514816284,
|
|
"epoch": 0.027763165585364426,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.000237,
|
|
"loss": 7.2908,
|
|
"mean_token_accuracy": 0.07796959653496742,
|
|
"num_tokens": 1030219.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.5679099559783936,
|
|
"epoch": 0.028055409433631422,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 7.3201,
|
|
"mean_token_accuracy": 0.08201361894607544,
|
|
"num_tokens": 1041911.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.6739061832427975,
|
|
"epoch": 0.028347653281898415,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000242,
|
|
"loss": 7.3237,
|
|
"mean_token_accuracy": 0.08364914655685425,
|
|
"num_tokens": 1052638.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.5074201107025145,
|
|
"epoch": 0.02863989713016541,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 7.2604,
|
|
"mean_token_accuracy": 0.08207505084574222,
|
|
"num_tokens": 1063465.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.632791471481323,
|
|
"epoch": 0.028932140978432403,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000247,
|
|
"loss": 7.2625,
|
|
"mean_token_accuracy": 0.08835876137018203,
|
|
"num_tokens": 1073583.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.605600214004516,
|
|
"epoch": 0.0292243848266994,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 7.327,
|
|
"mean_token_accuracy": 0.08249956965446473,
|
|
"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": 1602275401728000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|