2035 lines
54 KiB
JSON
2035 lines
54 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.0584487696533988,
|
|
"eval_steps": 500,
|
|
"global_step": 1000,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.691908264160157,
|
|
"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
|
|
},
|
|
{
|
|
"entropy": 7.432604742050171,
|
|
"epoch": 0.02951662867496639,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 0.000252,
|
|
"loss": 7.1818,
|
|
"mean_token_accuracy": 0.08461034670472145,
|
|
"num_tokens": 1095339.0,
|
|
"step": 505
|
|
},
|
|
{
|
|
"entropy": 7.534429121017456,
|
|
"epoch": 0.029808872523233387,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0002545,
|
|
"loss": 7.2615,
|
|
"mean_token_accuracy": 0.07991603463888168,
|
|
"num_tokens": 1105518.0,
|
|
"step": 510
|
|
},
|
|
{
|
|
"entropy": 7.572704219818116,
|
|
"epoch": 0.03010111637150038,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000257,
|
|
"loss": 7.1485,
|
|
"mean_token_accuracy": 0.08869004771113395,
|
|
"num_tokens": 1115666.0,
|
|
"step": 515
|
|
},
|
|
{
|
|
"entropy": 7.499062871932983,
|
|
"epoch": 0.030393360219767372,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0002595,
|
|
"loss": 7.2854,
|
|
"mean_token_accuracy": 0.08312782868742943,
|
|
"num_tokens": 1125827.0,
|
|
"step": 520
|
|
},
|
|
{
|
|
"entropy": 7.49239444732666,
|
|
"epoch": 0.030685604068034368,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.000262,
|
|
"loss": 7.2312,
|
|
"mean_token_accuracy": 0.08309667333960533,
|
|
"num_tokens": 1136154.0,
|
|
"step": 525
|
|
},
|
|
{
|
|
"entropy": 7.477995872497559,
|
|
"epoch": 0.03097784791630136,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00026450000000000003,
|
|
"loss": 7.2247,
|
|
"mean_token_accuracy": 0.09041514843702317,
|
|
"num_tokens": 1146392.0,
|
|
"step": 530
|
|
},
|
|
{
|
|
"entropy": 7.515807485580444,
|
|
"epoch": 0.03127009176456835,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00026700000000000004,
|
|
"loss": 7.2327,
|
|
"mean_token_accuracy": 0.08347614929080009,
|
|
"num_tokens": 1157566.0,
|
|
"step": 535
|
|
},
|
|
{
|
|
"entropy": 7.408181762695312,
|
|
"epoch": 0.03156233561283535,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.00026950000000000005,
|
|
"loss": 7.2244,
|
|
"mean_token_accuracy": 0.08636801540851594,
|
|
"num_tokens": 1167795.0,
|
|
"step": 540
|
|
},
|
|
{
|
|
"entropy": 7.55600643157959,
|
|
"epoch": 0.031854579461102345,
|
|
"grad_norm": 1.703125,
|
|
"learning_rate": 0.00027200000000000005,
|
|
"loss": 7.1614,
|
|
"mean_token_accuracy": 0.09404403418302536,
|
|
"num_tokens": 1178894.0,
|
|
"step": 545
|
|
},
|
|
{
|
|
"entropy": 7.379703378677368,
|
|
"epoch": 0.03214682330936934,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.0002745,
|
|
"loss": 7.1863,
|
|
"mean_token_accuracy": 0.08515683636069298,
|
|
"num_tokens": 1189361.0,
|
|
"step": 550
|
|
},
|
|
{
|
|
"entropy": 7.44665265083313,
|
|
"epoch": 0.03243906715763633,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.000277,
|
|
"loss": 7.2063,
|
|
"mean_token_accuracy": 0.08360610157251358,
|
|
"num_tokens": 1198992.0,
|
|
"step": 555
|
|
},
|
|
{
|
|
"entropy": 7.450517177581787,
|
|
"epoch": 0.032731311005903325,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.0002795,
|
|
"loss": 7.1858,
|
|
"mean_token_accuracy": 0.08214344829320908,
|
|
"num_tokens": 1208732.0,
|
|
"step": 560
|
|
},
|
|
{
|
|
"entropy": 7.416987085342408,
|
|
"epoch": 0.03302355485417032,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.00028199999999999997,
|
|
"loss": 7.1961,
|
|
"mean_token_accuracy": 0.08448583483695984,
|
|
"num_tokens": 1219015.0,
|
|
"step": 565
|
|
},
|
|
{
|
|
"entropy": 7.383893156051636,
|
|
"epoch": 0.03331579870243731,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.0002845,
|
|
"loss": 7.0907,
|
|
"mean_token_accuracy": 0.10338108763098716,
|
|
"num_tokens": 1229227.0,
|
|
"step": 570
|
|
},
|
|
{
|
|
"entropy": 7.4993305683135985,
|
|
"epoch": 0.033608042550704306,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.000287,
|
|
"loss": 7.2226,
|
|
"mean_token_accuracy": 0.08231785595417022,
|
|
"num_tokens": 1239654.0,
|
|
"step": 575
|
|
},
|
|
{
|
|
"entropy": 7.332780599594116,
|
|
"epoch": 0.0339002863989713,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0002895,
|
|
"loss": 7.1941,
|
|
"mean_token_accuracy": 0.08100020363926888,
|
|
"num_tokens": 1250955.0,
|
|
"step": 580
|
|
},
|
|
{
|
|
"entropy": 7.445628499984741,
|
|
"epoch": 0.0341925302472383,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000292,
|
|
"loss": 7.1521,
|
|
"mean_token_accuracy": 0.09030932411551476,
|
|
"num_tokens": 1262574.0,
|
|
"step": 585
|
|
},
|
|
{
|
|
"entropy": 7.327143812179566,
|
|
"epoch": 0.03448477409550529,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0002945,
|
|
"loss": 7.1163,
|
|
"mean_token_accuracy": 0.09456900656223297,
|
|
"num_tokens": 1273272.0,
|
|
"step": 590
|
|
},
|
|
{
|
|
"entropy": 7.291459846496582,
|
|
"epoch": 0.03477701794377228,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000297,
|
|
"loss": 7.1011,
|
|
"mean_token_accuracy": 0.09272942468523979,
|
|
"num_tokens": 1284001.0,
|
|
"step": 595
|
|
},
|
|
{
|
|
"entropy": 7.443050575256348,
|
|
"epoch": 0.03506926179203928,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.0002995,
|
|
"loss": 7.1401,
|
|
"mean_token_accuracy": 0.08779146820306778,
|
|
"num_tokens": 1294671.0,
|
|
"step": 600
|
|
},
|
|
{
|
|
"entropy": 7.328179550170899,
|
|
"epoch": 0.035361505640306275,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000302,
|
|
"loss": 7.1451,
|
|
"mean_token_accuracy": 0.09261744394898415,
|
|
"num_tokens": 1305397.0,
|
|
"step": 605
|
|
},
|
|
{
|
|
"entropy": 7.390170431137085,
|
|
"epoch": 0.035653749488573264,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0003045,
|
|
"loss": 7.1696,
|
|
"mean_token_accuracy": 0.08689776062965393,
|
|
"num_tokens": 1317449.0,
|
|
"step": 610
|
|
},
|
|
{
|
|
"entropy": 7.28363561630249,
|
|
"epoch": 0.03594599333684026,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.000307,
|
|
"loss": 7.0866,
|
|
"mean_token_accuracy": 0.08486035652458668,
|
|
"num_tokens": 1328645.0,
|
|
"step": 615
|
|
},
|
|
{
|
|
"entropy": 7.2835465431213375,
|
|
"epoch": 0.036238237185107255,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0003095,
|
|
"loss": 7.091,
|
|
"mean_token_accuracy": 0.0892544224858284,
|
|
"num_tokens": 1339815.0,
|
|
"step": 620
|
|
},
|
|
{
|
|
"entropy": 7.290801572799682,
|
|
"epoch": 0.036530481033374244,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.000312,
|
|
"loss": 7.0954,
|
|
"mean_token_accuracy": 0.08872693330049515,
|
|
"num_tokens": 1350301.0,
|
|
"step": 625
|
|
},
|
|
{
|
|
"entropy": 7.280572080612183,
|
|
"epoch": 0.03682272488164124,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0003145,
|
|
"loss": 7.0829,
|
|
"mean_token_accuracy": 0.09459017589688301,
|
|
"num_tokens": 1360928.0,
|
|
"step": 630
|
|
},
|
|
{
|
|
"entropy": 7.34552264213562,
|
|
"epoch": 0.037114968729908236,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000317,
|
|
"loss": 7.1047,
|
|
"mean_token_accuracy": 0.08707956597208977,
|
|
"num_tokens": 1371618.0,
|
|
"step": 635
|
|
},
|
|
{
|
|
"entropy": 7.211892175674438,
|
|
"epoch": 0.03740721257817523,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0003195,
|
|
"loss": 7.0702,
|
|
"mean_token_accuracy": 0.0876346081495285,
|
|
"num_tokens": 1382824.0,
|
|
"step": 640
|
|
},
|
|
{
|
|
"entropy": 7.2274785995483395,
|
|
"epoch": 0.03769945642644222,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.000322,
|
|
"loss": 7.0401,
|
|
"mean_token_accuracy": 0.09252982586622238,
|
|
"num_tokens": 1393244.0,
|
|
"step": 645
|
|
},
|
|
{
|
|
"entropy": 7.224736738204956,
|
|
"epoch": 0.03799170027470922,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00032450000000000003,
|
|
"loss": 7.0136,
|
|
"mean_token_accuracy": 0.09895449355244637,
|
|
"num_tokens": 1403469.0,
|
|
"step": 650
|
|
},
|
|
{
|
|
"entropy": 7.358276271820069,
|
|
"epoch": 0.03828394412297621,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.00032700000000000003,
|
|
"loss": 7.086,
|
|
"mean_token_accuracy": 0.09135140627622604,
|
|
"num_tokens": 1413860.0,
|
|
"step": 655
|
|
},
|
|
{
|
|
"entropy": 7.256574583053589,
|
|
"epoch": 0.03857618797124321,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.00032950000000000004,
|
|
"loss": 7.1348,
|
|
"mean_token_accuracy": 0.08090706467628479,
|
|
"num_tokens": 1423985.0,
|
|
"step": 660
|
|
},
|
|
{
|
|
"entropy": 7.278745222091675,
|
|
"epoch": 0.0388684318195102,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.00033200000000000005,
|
|
"loss": 7.087,
|
|
"mean_token_accuracy": 0.09359103664755822,
|
|
"num_tokens": 1434838.0,
|
|
"step": 665
|
|
},
|
|
{
|
|
"entropy": 7.2903660297393795,
|
|
"epoch": 0.039160675667777194,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.00033450000000000005,
|
|
"loss": 7.007,
|
|
"mean_token_accuracy": 0.08955025300383568,
|
|
"num_tokens": 1446429.0,
|
|
"step": 670
|
|
},
|
|
{
|
|
"entropy": 7.332035541534424,
|
|
"epoch": 0.03945291951604419,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.000337,
|
|
"loss": 7.1183,
|
|
"mean_token_accuracy": 0.08762314394116402,
|
|
"num_tokens": 1456726.0,
|
|
"step": 675
|
|
},
|
|
{
|
|
"entropy": 7.197989416122437,
|
|
"epoch": 0.03974516336431118,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.0003395,
|
|
"loss": 7.0058,
|
|
"mean_token_accuracy": 0.09932290315628052,
|
|
"num_tokens": 1466378.0,
|
|
"step": 680
|
|
},
|
|
{
|
|
"entropy": 7.104801225662231,
|
|
"epoch": 0.040037407212578174,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 0.000342,
|
|
"loss": 6.9835,
|
|
"mean_token_accuracy": 0.09476312026381492,
|
|
"num_tokens": 1477755.0,
|
|
"step": 685
|
|
},
|
|
{
|
|
"entropy": 7.237457656860352,
|
|
"epoch": 0.04032965106084517,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.00034449999999999997,
|
|
"loss": 7.036,
|
|
"mean_token_accuracy": 0.08923429250717163,
|
|
"num_tokens": 1488209.0,
|
|
"step": 690
|
|
},
|
|
{
|
|
"entropy": 7.137836837768555,
|
|
"epoch": 0.040621894909112166,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.000347,
|
|
"loss": 7.0728,
|
|
"mean_token_accuracy": 0.08890949785709382,
|
|
"num_tokens": 1498658.0,
|
|
"step": 695
|
|
},
|
|
{
|
|
"entropy": 7.236037731170654,
|
|
"epoch": 0.040914138757379155,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0003495,
|
|
"loss": 7.0789,
|
|
"mean_token_accuracy": 0.0860823355615139,
|
|
"num_tokens": 1509766.0,
|
|
"step": 700
|
|
},
|
|
{
|
|
"entropy": 7.218784284591675,
|
|
"epoch": 0.04120638260564615,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 0.000352,
|
|
"loss": 6.975,
|
|
"mean_token_accuracy": 0.0937421977519989,
|
|
"num_tokens": 1521139.0,
|
|
"step": 705
|
|
},
|
|
{
|
|
"entropy": 7.174445295333863,
|
|
"epoch": 0.04149862645391315,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.0003545,
|
|
"loss": 7.0357,
|
|
"mean_token_accuracy": 0.0900140330195427,
|
|
"num_tokens": 1531875.0,
|
|
"step": 710
|
|
},
|
|
{
|
|
"entropy": 7.153319358825684,
|
|
"epoch": 0.041790870302180136,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.000357,
|
|
"loss": 7.0205,
|
|
"mean_token_accuracy": 0.08912704512476921,
|
|
"num_tokens": 1542678.0,
|
|
"step": 715
|
|
},
|
|
{
|
|
"entropy": 7.196989250183106,
|
|
"epoch": 0.04208311415044713,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.0003595,
|
|
"loss": 6.9703,
|
|
"mean_token_accuracy": 0.09389843866229057,
|
|
"num_tokens": 1553882.0,
|
|
"step": 720
|
|
},
|
|
{
|
|
"entropy": 7.12842378616333,
|
|
"epoch": 0.04237535799871413,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000362,
|
|
"loss": 7.0568,
|
|
"mean_token_accuracy": 0.0915483608841896,
|
|
"num_tokens": 1564854.0,
|
|
"step": 725
|
|
},
|
|
{
|
|
"entropy": 7.176506567001343,
|
|
"epoch": 0.042667601846981124,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0003645,
|
|
"loss": 6.9303,
|
|
"mean_token_accuracy": 0.09940993562340736,
|
|
"num_tokens": 1576112.0,
|
|
"step": 730
|
|
},
|
|
{
|
|
"entropy": 7.137564849853516,
|
|
"epoch": 0.04295984569524811,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.000367,
|
|
"loss": 6.9771,
|
|
"mean_token_accuracy": 0.0963865041732788,
|
|
"num_tokens": 1586235.0,
|
|
"step": 735
|
|
},
|
|
{
|
|
"entropy": 7.083181285858155,
|
|
"epoch": 0.04325208954351511,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 0.0003695,
|
|
"loss": 6.9769,
|
|
"mean_token_accuracy": 0.09416570290923118,
|
|
"num_tokens": 1596727.0,
|
|
"step": 740
|
|
},
|
|
{
|
|
"entropy": 7.099421977996826,
|
|
"epoch": 0.043544333391782104,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000372,
|
|
"loss": 6.8926,
|
|
"mean_token_accuracy": 0.09487500786781311,
|
|
"num_tokens": 1607008.0,
|
|
"step": 745
|
|
},
|
|
{
|
|
"entropy": 7.178588104248047,
|
|
"epoch": 0.0438365772400491,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.0003745,
|
|
"loss": 6.9975,
|
|
"mean_token_accuracy": 0.0891700565814972,
|
|
"num_tokens": 1618081.0,
|
|
"step": 750
|
|
},
|
|
{
|
|
"entropy": 7.064407539367676,
|
|
"epoch": 0.04412882108831609,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000377,
|
|
"loss": 6.9569,
|
|
"mean_token_accuracy": 0.08645836561918259,
|
|
"num_tokens": 1629095.0,
|
|
"step": 755
|
|
},
|
|
{
|
|
"entropy": 7.101617288589478,
|
|
"epoch": 0.044421064936583085,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0003795,
|
|
"loss": 6.9537,
|
|
"mean_token_accuracy": 0.0951661191880703,
|
|
"num_tokens": 1639861.0,
|
|
"step": 760
|
|
},
|
|
{
|
|
"entropy": 7.189848804473877,
|
|
"epoch": 0.04471330878485008,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.000382,
|
|
"loss": 7.0003,
|
|
"mean_token_accuracy": 0.09473721534013749,
|
|
"num_tokens": 1649169.0,
|
|
"step": 765
|
|
},
|
|
{
|
|
"entropy": 6.957789611816406,
|
|
"epoch": 0.04500555263311707,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0003845,
|
|
"loss": 6.8833,
|
|
"mean_token_accuracy": 0.10007616281509399,
|
|
"num_tokens": 1658528.0,
|
|
"step": 770
|
|
},
|
|
{
|
|
"entropy": 7.049975156784058,
|
|
"epoch": 0.045297796481384066,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00038700000000000003,
|
|
"loss": 6.9657,
|
|
"mean_token_accuracy": 0.09249654635787011,
|
|
"num_tokens": 1669231.0,
|
|
"step": 775
|
|
},
|
|
{
|
|
"entropy": 7.0429846286773685,
|
|
"epoch": 0.04559004032965106,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.00038950000000000003,
|
|
"loss": 6.8987,
|
|
"mean_token_accuracy": 0.09256255626678467,
|
|
"num_tokens": 1680503.0,
|
|
"step": 780
|
|
},
|
|
{
|
|
"entropy": 7.069688749313355,
|
|
"epoch": 0.04588228417791806,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.00039200000000000004,
|
|
"loss": 6.9239,
|
|
"mean_token_accuracy": 0.09164713248610497,
|
|
"num_tokens": 1690827.0,
|
|
"step": 785
|
|
},
|
|
{
|
|
"entropy": 7.1588634014129635,
|
|
"epoch": 0.04617452802618505,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.00039450000000000005,
|
|
"loss": 6.992,
|
|
"mean_token_accuracy": 0.09126574769616128,
|
|
"num_tokens": 1702931.0,
|
|
"step": 790
|
|
},
|
|
{
|
|
"entropy": 6.960534286499024,
|
|
"epoch": 0.04646677187445204,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00039700000000000005,
|
|
"loss": 6.8832,
|
|
"mean_token_accuracy": 0.09681101739406586,
|
|
"num_tokens": 1712834.0,
|
|
"step": 795
|
|
},
|
|
{
|
|
"entropy": 7.059118413925171,
|
|
"epoch": 0.04675901572271904,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0003995,
|
|
"loss": 6.9505,
|
|
"mean_token_accuracy": 0.09173840507864953,
|
|
"num_tokens": 1723289.0,
|
|
"step": 800
|
|
},
|
|
{
|
|
"entropy": 6.98992805480957,
|
|
"epoch": 0.04705125957098603,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000402,
|
|
"loss": 6.9111,
|
|
"mean_token_accuracy": 0.09471877589821816,
|
|
"num_tokens": 1734477.0,
|
|
"step": 805
|
|
},
|
|
{
|
|
"entropy": 7.110216426849365,
|
|
"epoch": 0.04734350341925302,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0004045,
|
|
"loss": 6.9201,
|
|
"mean_token_accuracy": 0.09181275144219399,
|
|
"num_tokens": 1745187.0,
|
|
"step": 810
|
|
},
|
|
{
|
|
"entropy": 6.9768541812896725,
|
|
"epoch": 0.04763574726752002,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.00040699999999999997,
|
|
"loss": 6.8563,
|
|
"mean_token_accuracy": 0.09723289534449578,
|
|
"num_tokens": 1756687.0,
|
|
"step": 815
|
|
},
|
|
{
|
|
"entropy": 7.012206411361694,
|
|
"epoch": 0.047927991115787015,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0004095,
|
|
"loss": 6.9795,
|
|
"mean_token_accuracy": 0.09605197682976722,
|
|
"num_tokens": 1767928.0,
|
|
"step": 820
|
|
},
|
|
{
|
|
"entropy": 7.063567161560059,
|
|
"epoch": 0.048220234964054004,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000412,
|
|
"loss": 6.9206,
|
|
"mean_token_accuracy": 0.0963786244392395,
|
|
"num_tokens": 1777951.0,
|
|
"step": 825
|
|
},
|
|
{
|
|
"entropy": 6.963260746002197,
|
|
"epoch": 0.048512478812321,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0004145,
|
|
"loss": 6.9047,
|
|
"mean_token_accuracy": 0.09518637284636497,
|
|
"num_tokens": 1788598.0,
|
|
"step": 830
|
|
},
|
|
{
|
|
"entropy": 7.022843551635742,
|
|
"epoch": 0.048804722660587996,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000417,
|
|
"loss": 6.8708,
|
|
"mean_token_accuracy": 0.0998458556830883,
|
|
"num_tokens": 1799446.0,
|
|
"step": 835
|
|
},
|
|
{
|
|
"entropy": 7.028498888015747,
|
|
"epoch": 0.04909696650885499,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.0004195,
|
|
"loss": 6.8757,
|
|
"mean_token_accuracy": 0.0957573838531971,
|
|
"num_tokens": 1809687.0,
|
|
"step": 840
|
|
},
|
|
{
|
|
"entropy": 6.976888608932495,
|
|
"epoch": 0.04938921035712198,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.000422,
|
|
"loss": 6.9107,
|
|
"mean_token_accuracy": 0.09788398742675782,
|
|
"num_tokens": 1820535.0,
|
|
"step": 845
|
|
},
|
|
{
|
|
"entropy": 6.943860197067261,
|
|
"epoch": 0.04968145420538898,
|
|
"grad_norm": 1.6640625,
|
|
"learning_rate": 0.0004245,
|
|
"loss": 6.8376,
|
|
"mean_token_accuracy": 0.09310686662793159,
|
|
"num_tokens": 1830834.0,
|
|
"step": 850
|
|
},
|
|
{
|
|
"entropy": 6.961981105804443,
|
|
"epoch": 0.04997369805365597,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000427,
|
|
"loss": 6.8257,
|
|
"mean_token_accuracy": 0.0944649763405323,
|
|
"num_tokens": 1842615.0,
|
|
"step": 855
|
|
},
|
|
{
|
|
"entropy": 6.919750738143921,
|
|
"epoch": 0.05026594190192296,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0004295,
|
|
"loss": 6.8328,
|
|
"mean_token_accuracy": 0.09815948903560638,
|
|
"num_tokens": 1852844.0,
|
|
"step": 860
|
|
},
|
|
{
|
|
"entropy": 6.876930141448975,
|
|
"epoch": 0.05055818575018996,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000432,
|
|
"loss": 6.8918,
|
|
"mean_token_accuracy": 0.09418935924768448,
|
|
"num_tokens": 1862528.0,
|
|
"step": 865
|
|
},
|
|
{
|
|
"entropy": 6.934880113601684,
|
|
"epoch": 0.05085042959845695,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.0004345,
|
|
"loss": 6.8469,
|
|
"mean_token_accuracy": 0.10216829851269722,
|
|
"num_tokens": 1873275.0,
|
|
"step": 870
|
|
},
|
|
{
|
|
"entropy": 6.956448936462403,
|
|
"epoch": 0.05114267344672395,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000437,
|
|
"loss": 6.8847,
|
|
"mean_token_accuracy": 0.09507919698953629,
|
|
"num_tokens": 1884724.0,
|
|
"step": 875
|
|
},
|
|
{
|
|
"entropy": 7.051344537734986,
|
|
"epoch": 0.05143491729499094,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.0004395,
|
|
"loss": 6.8598,
|
|
"mean_token_accuracy": 0.09828044772148133,
|
|
"num_tokens": 1894705.0,
|
|
"step": 880
|
|
},
|
|
{
|
|
"entropy": 6.862812566757202,
|
|
"epoch": 0.051727161143257934,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.000442,
|
|
"loss": 6.9184,
|
|
"mean_token_accuracy": 0.09541104733943939,
|
|
"num_tokens": 1906553.0,
|
|
"step": 885
|
|
},
|
|
{
|
|
"entropy": 7.020113182067871,
|
|
"epoch": 0.05201940499152493,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0004445,
|
|
"loss": 6.8544,
|
|
"mean_token_accuracy": 0.10240027904510499,
|
|
"num_tokens": 1916628.0,
|
|
"step": 890
|
|
},
|
|
{
|
|
"entropy": 6.766639471054077,
|
|
"epoch": 0.052311648839791926,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.000447,
|
|
"loss": 6.6849,
|
|
"mean_token_accuracy": 0.11086282283067703,
|
|
"num_tokens": 1926686.0,
|
|
"step": 895
|
|
},
|
|
{
|
|
"entropy": 6.947674942016602,
|
|
"epoch": 0.052603892688058915,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 0.00044950000000000003,
|
|
"loss": 6.8638,
|
|
"mean_token_accuracy": 0.09778930097818375,
|
|
"num_tokens": 1938150.0,
|
|
"step": 900
|
|
},
|
|
{
|
|
"entropy": 6.891585969924927,
|
|
"epoch": 0.05289613653632591,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.00045200000000000004,
|
|
"loss": 6.873,
|
|
"mean_token_accuracy": 0.09291322231292724,
|
|
"num_tokens": 1948617.0,
|
|
"step": 905
|
|
},
|
|
{
|
|
"entropy": 6.900164985656739,
|
|
"epoch": 0.053188380384592907,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00045450000000000004,
|
|
"loss": 6.8565,
|
|
"mean_token_accuracy": 0.09954984486103058,
|
|
"num_tokens": 1958943.0,
|
|
"step": 910
|
|
},
|
|
{
|
|
"entropy": 6.856646156311035,
|
|
"epoch": 0.053480624232859895,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.00045700000000000005,
|
|
"loss": 6.7881,
|
|
"mean_token_accuracy": 0.10126558542251587,
|
|
"num_tokens": 1968988.0,
|
|
"step": 915
|
|
},
|
|
{
|
|
"entropy": 6.895283889770508,
|
|
"epoch": 0.05377286808112689,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.00045950000000000006,
|
|
"loss": 6.7692,
|
|
"mean_token_accuracy": 0.10075787082314491,
|
|
"num_tokens": 1979490.0,
|
|
"step": 920
|
|
},
|
|
{
|
|
"entropy": 6.938627767562866,
|
|
"epoch": 0.05406511192939389,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.000462,
|
|
"loss": 6.8187,
|
|
"mean_token_accuracy": 0.10156044065952301,
|
|
"num_tokens": 1990377.0,
|
|
"step": 925
|
|
},
|
|
{
|
|
"entropy": 6.881887578964234,
|
|
"epoch": 0.05435735577766088,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0004645,
|
|
"loss": 6.7914,
|
|
"mean_token_accuracy": 0.09942576885223389,
|
|
"num_tokens": 2001678.0,
|
|
"step": 930
|
|
},
|
|
{
|
|
"entropy": 6.809711265563965,
|
|
"epoch": 0.05464959962592787,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000467,
|
|
"loss": 6.7898,
|
|
"mean_token_accuracy": 0.11002987027168273,
|
|
"num_tokens": 2011550.0,
|
|
"step": 935
|
|
},
|
|
{
|
|
"entropy": 6.853000497817993,
|
|
"epoch": 0.05494184347419487,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0004695,
|
|
"loss": 6.8297,
|
|
"mean_token_accuracy": 0.10066121965646743,
|
|
"num_tokens": 2021607.0,
|
|
"step": 940
|
|
},
|
|
{
|
|
"entropy": 6.862749242782593,
|
|
"epoch": 0.055234087322461864,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000472,
|
|
"loss": 6.8211,
|
|
"mean_token_accuracy": 0.09814092963933944,
|
|
"num_tokens": 2030975.0,
|
|
"step": 945
|
|
},
|
|
{
|
|
"entropy": 6.836636781692505,
|
|
"epoch": 0.05552633117072885,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0004745,
|
|
"loss": 6.8844,
|
|
"mean_token_accuracy": 0.09624653086066245,
|
|
"num_tokens": 2041799.0,
|
|
"step": 950
|
|
},
|
|
{
|
|
"entropy": 6.891287755966187,
|
|
"epoch": 0.05581857501899585,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000477,
|
|
"loss": 6.7935,
|
|
"mean_token_accuracy": 0.10220010504126549,
|
|
"num_tokens": 2052960.0,
|
|
"step": 955
|
|
},
|
|
{
|
|
"entropy": 6.868294525146484,
|
|
"epoch": 0.056110818867262845,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0004795,
|
|
"loss": 6.8472,
|
|
"mean_token_accuracy": 0.09924568086862565,
|
|
"num_tokens": 2064156.0,
|
|
"step": 960
|
|
},
|
|
{
|
|
"entropy": 6.914956092834473,
|
|
"epoch": 0.05640306271552984,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000482,
|
|
"loss": 6.8016,
|
|
"mean_token_accuracy": 0.100481665879488,
|
|
"num_tokens": 2074774.0,
|
|
"step": 965
|
|
},
|
|
{
|
|
"entropy": 6.793342113494873,
|
|
"epoch": 0.05669530656379683,
|
|
"grad_norm": 0.9140625,
|
|
"learning_rate": 0.0004845,
|
|
"loss": 6.8288,
|
|
"mean_token_accuracy": 0.09989185482263566,
|
|
"num_tokens": 2086164.0,
|
|
"step": 970
|
|
},
|
|
{
|
|
"entropy": 6.78997688293457,
|
|
"epoch": 0.056987550412063825,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.000487,
|
|
"loss": 6.6708,
|
|
"mean_token_accuracy": 0.1248696506023407,
|
|
"num_tokens": 2096268.0,
|
|
"step": 975
|
|
},
|
|
{
|
|
"entropy": 6.888183069229126,
|
|
"epoch": 0.05727979426033082,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.0004895,
|
|
"loss": 6.8985,
|
|
"mean_token_accuracy": 0.09069314152002335,
|
|
"num_tokens": 2108511.0,
|
|
"step": 980
|
|
},
|
|
{
|
|
"entropy": 6.923859405517578,
|
|
"epoch": 0.05757203810859782,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000492,
|
|
"loss": 6.8155,
|
|
"mean_token_accuracy": 0.09430013298988342,
|
|
"num_tokens": 2119622.0,
|
|
"step": 985
|
|
},
|
|
{
|
|
"entropy": 6.742640447616577,
|
|
"epoch": 0.057864281956864806,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0004945,
|
|
"loss": 6.8703,
|
|
"mean_token_accuracy": 0.10323725119233132,
|
|
"num_tokens": 2131021.0,
|
|
"step": 990
|
|
},
|
|
{
|
|
"entropy": 6.867950057983398,
|
|
"epoch": 0.0581565258051318,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000497,
|
|
"loss": 6.8448,
|
|
"mean_token_accuracy": 0.10064574256539345,
|
|
"num_tokens": 2141797.0,
|
|
"step": 995
|
|
},
|
|
{
|
|
"entropy": 6.782402372360229,
|
|
"epoch": 0.0584487696533988,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.0004995,
|
|
"loss": 6.7141,
|
|
"mean_token_accuracy": 0.09859876409173012,
|
|
"num_tokens": 2153609.0,
|
|
"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": 3212952975360000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|