1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.03217917363882095,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.742555713653564,
|
|
"epoch": 0.00032179173638820956,
|
|
"grad_norm": 4.59375,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7932,
|
|
"mean_token_accuracy": 9.85221704468131e-05,
|
|
"num_tokens": 11506.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.742585468292237,
|
|
"epoch": 0.0006435834727764191,
|
|
"grad_norm": 5.0,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7664,
|
|
"mean_token_accuracy": 7.942811935208738e-05,
|
|
"num_tokens": 21973.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.742585182189941,
|
|
"epoch": 0.0009653752091646286,
|
|
"grad_norm": 4.6875,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.7473,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 32763.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.742647171020508,
|
|
"epoch": 0.0012871669455528382,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.716,
|
|
"mean_token_accuracy": 0.00042943707667291163,
|
|
"num_tokens": 44354.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.742597675323486,
|
|
"epoch": 0.0016089586819410479,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.6219,
|
|
"mean_token_accuracy": 0.0008890067110769451,
|
|
"num_tokens": 56534.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.74241762161255,
|
|
"epoch": 0.0019307504183292573,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.4911,
|
|
"mean_token_accuracy": 0.009375615860335529,
|
|
"num_tokens": 67106.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.741784954071045,
|
|
"epoch": 0.002252542154717467,
|
|
"grad_norm": 3.265625,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.3831,
|
|
"mean_token_accuracy": 0.021833056677132846,
|
|
"num_tokens": 78801.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.74044713973999,
|
|
"epoch": 0.0025743338911056765,
|
|
"grad_norm": 2.53125,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.2405,
|
|
"mean_token_accuracy": 0.028899907879531384,
|
|
"num_tokens": 90390.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.738579750061035,
|
|
"epoch": 0.002896125627493886,
|
|
"grad_norm": 2.265625,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.1128,
|
|
"mean_token_accuracy": 0.036285107396543025,
|
|
"num_tokens": 101652.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.737388038635254,
|
|
"epoch": 0.0032179173638820957,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 10.0384,
|
|
"mean_token_accuracy": 0.03730290364474058,
|
|
"num_tokens": 112068.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.737480354309081,
|
|
"epoch": 0.003539709100270305,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 10.0098,
|
|
"mean_token_accuracy": 0.04103992246091366,
|
|
"num_tokens": 123500.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.736884212493896,
|
|
"epoch": 0.0038615008366585145,
|
|
"grad_norm": 1.9296875,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.9544,
|
|
"mean_token_accuracy": 0.0359757786616683,
|
|
"num_tokens": 134456.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.7358078956604,
|
|
"epoch": 0.004183292573046725,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.9017,
|
|
"mean_token_accuracy": 0.03450035620480776,
|
|
"num_tokens": 146299.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.734182739257813,
|
|
"epoch": 0.004505084309434934,
|
|
"grad_norm": 1.9375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.8076,
|
|
"mean_token_accuracy": 0.037839466519653794,
|
|
"num_tokens": 157494.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.73177366256714,
|
|
"epoch": 0.004826876045823143,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.7775,
|
|
"mean_token_accuracy": 0.03611580468714237,
|
|
"num_tokens": 168226.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.728845882415772,
|
|
"epoch": 0.005148667782211353,
|
|
"grad_norm": 1.9296875,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.6556,
|
|
"mean_token_accuracy": 0.04190883524715901,
|
|
"num_tokens": 179311.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.725195121765136,
|
|
"epoch": 0.005470459518599562,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.6267,
|
|
"mean_token_accuracy": 0.043489571660757065,
|
|
"num_tokens": 191513.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.719332885742187,
|
|
"epoch": 0.005792251254987772,
|
|
"grad_norm": 1.8828125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.5397,
|
|
"mean_token_accuracy": 0.046276621893048285,
|
|
"num_tokens": 204631.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.710766220092774,
|
|
"epoch": 0.006114042991375981,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.4486,
|
|
"mean_token_accuracy": 0.0460438147187233,
|
|
"num_tokens": 215901.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.69977626800537,
|
|
"epoch": 0.0064358347277641914,
|
|
"grad_norm": 1.8203125,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.4013,
|
|
"mean_token_accuracy": 0.044720832630991934,
|
|
"num_tokens": 228928.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.685317134857177,
|
|
"epoch": 0.006757626464152401,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.3138,
|
|
"mean_token_accuracy": 0.044154060259461406,
|
|
"num_tokens": 241543.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.66606101989746,
|
|
"epoch": 0.00707941820054061,
|
|
"grad_norm": 1.8203125,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.1466,
|
|
"mean_token_accuracy": 0.046302244812250135,
|
|
"num_tokens": 253267.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.636186695098877,
|
|
"epoch": 0.00740120993692882,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 9.0141,
|
|
"mean_token_accuracy": 0.05802988223731518,
|
|
"num_tokens": 264760.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.5947696685791,
|
|
"epoch": 0.007723001673317029,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.9075,
|
|
"mean_token_accuracy": 0.051640312373638156,
|
|
"num_tokens": 275080.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.542396450042725,
|
|
"epoch": 0.008044793409705238,
|
|
"grad_norm": 1.7109375,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.7919,
|
|
"mean_token_accuracy": 0.05795170105993748,
|
|
"num_tokens": 287114.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.490421962738036,
|
|
"epoch": 0.00836658514609345,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.6367,
|
|
"mean_token_accuracy": 0.06025359183549881,
|
|
"num_tokens": 298732.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.417353057861328,
|
|
"epoch": 0.008688376882481658,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.4641,
|
|
"mean_token_accuracy": 0.06998355574905872,
|
|
"num_tokens": 310227.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.32230224609375,
|
|
"epoch": 0.009010168618869868,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.3943,
|
|
"mean_token_accuracy": 0.06697470992803574,
|
|
"num_tokens": 322724.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.228569889068604,
|
|
"epoch": 0.009331960355258077,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.3137,
|
|
"mean_token_accuracy": 0.06024596691131592,
|
|
"num_tokens": 335638.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.116732311248779,
|
|
"epoch": 0.009653752091646286,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.1548,
|
|
"mean_token_accuracy": 0.07251053638756275,
|
|
"num_tokens": 346751.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.953464126586914,
|
|
"epoch": 0.009975543828034497,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.0894,
|
|
"mean_token_accuracy": 0.07281172573566437,
|
|
"num_tokens": 357809.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.77461576461792,
|
|
"epoch": 0.010297335564422706,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.9142,
|
|
"mean_token_accuracy": 0.06864561475813388,
|
|
"num_tokens": 369631.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.603110790252686,
|
|
"epoch": 0.010619127300810915,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.9575,
|
|
"mean_token_accuracy": 0.06908577904105187,
|
|
"num_tokens": 380904.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.377412414550781,
|
|
"epoch": 0.010940919037199124,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.7359,
|
|
"mean_token_accuracy": 0.07545260861515998,
|
|
"num_tokens": 392898.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.168265438079834,
|
|
"epoch": 0.011262710773587334,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.6702,
|
|
"mean_token_accuracy": 0.07459548935294151,
|
|
"num_tokens": 404285.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.926816463470459,
|
|
"epoch": 0.011584502509975544,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.5717,
|
|
"mean_token_accuracy": 0.07761912979185581,
|
|
"num_tokens": 416443.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.745314025878907,
|
|
"epoch": 0.011906294246363754,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.6426,
|
|
"mean_token_accuracy": 0.0765433344990015,
|
|
"num_tokens": 426558.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.587053394317627,
|
|
"epoch": 0.012228085982751963,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.6648,
|
|
"mean_token_accuracy": 0.07643079459667206,
|
|
"num_tokens": 437794.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.517185878753661,
|
|
"epoch": 0.012549877719140172,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.4958,
|
|
"mean_token_accuracy": 0.07788795456290246,
|
|
"num_tokens": 448682.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.376048374176026,
|
|
"epoch": 0.012871669455528383,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.417,
|
|
"mean_token_accuracy": 0.08175441175699234,
|
|
"num_tokens": 459699.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.276328945159912,
|
|
"epoch": 0.013193461191916592,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.3537,
|
|
"mean_token_accuracy": 0.08016520962119103,
|
|
"num_tokens": 472857.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.274100303649902,
|
|
"epoch": 0.013515252928304801,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.4194,
|
|
"mean_token_accuracy": 0.08162855952978135,
|
|
"num_tokens": 484983.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.180224132537841,
|
|
"epoch": 0.01383704466469301,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.3626,
|
|
"mean_token_accuracy": 0.08450362831354141,
|
|
"num_tokens": 495973.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.187409973144531,
|
|
"epoch": 0.01415883640108122,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.2816,
|
|
"mean_token_accuracy": 0.090892593562603,
|
|
"num_tokens": 507017.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.120441341400147,
|
|
"epoch": 0.01448062813746943,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.36,
|
|
"mean_token_accuracy": 0.08900276795029641,
|
|
"num_tokens": 517522.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.129059314727783,
|
|
"epoch": 0.01480241987385764,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.3701,
|
|
"mean_token_accuracy": 0.09131013453006745,
|
|
"num_tokens": 528240.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.994166135787964,
|
|
"epoch": 0.015124211610245849,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.328,
|
|
"mean_token_accuracy": 0.0881452701985836,
|
|
"num_tokens": 538875.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.10927906036377,
|
|
"epoch": 0.015446003346634058,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.3722,
|
|
"mean_token_accuracy": 0.08998035714030266,
|
|
"num_tokens": 549386.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 8.068589448928833,
|
|
"epoch": 0.015767795083022267,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.3029,
|
|
"mean_token_accuracy": 0.08936483487486839,
|
|
"num_tokens": 559643.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.049430751800537,
|
|
"epoch": 0.016089586819410476,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.2267,
|
|
"mean_token_accuracy": 0.09506414234638214,
|
|
"num_tokens": 570932.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.93134560585022,
|
|
"epoch": 0.016411378555798686,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.2019,
|
|
"mean_token_accuracy": 0.09725837931036949,
|
|
"num_tokens": 580258.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.893705558776856,
|
|
"epoch": 0.0167331702921869,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.1337,
|
|
"mean_token_accuracy": 0.09558385238051414,
|
|
"num_tokens": 590622.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.884620428085327,
|
|
"epoch": 0.017054962028575107,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.1827,
|
|
"mean_token_accuracy": 0.09120650067925454,
|
|
"num_tokens": 602202.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.823669528961181,
|
|
"epoch": 0.017376753764963317,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.1309,
|
|
"mean_token_accuracy": 0.0979801706969738,
|
|
"num_tokens": 613864.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.87017560005188,
|
|
"epoch": 0.017698545501351526,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.2524,
|
|
"mean_token_accuracy": 0.09466890171170235,
|
|
"num_tokens": 626884.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.857308626174927,
|
|
"epoch": 0.018020337237739735,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.1612,
|
|
"mean_token_accuracy": 0.09996400326490402,
|
|
"num_tokens": 638418.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.856085300445557,
|
|
"epoch": 0.018342128974127944,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.1939,
|
|
"mean_token_accuracy": 0.09103256016969681,
|
|
"num_tokens": 648376.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.9166460037231445,
|
|
"epoch": 0.018663920710516153,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.2464,
|
|
"mean_token_accuracy": 0.09389530643820762,
|
|
"num_tokens": 659268.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.788627672195434,
|
|
"epoch": 0.018985712446904363,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.0223,
|
|
"mean_token_accuracy": 0.10249627456068992,
|
|
"num_tokens": 669736.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.768741464614868,
|
|
"epoch": 0.01930750418329257,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.1198,
|
|
"mean_token_accuracy": 0.0995169997215271,
|
|
"num_tokens": 681571.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.723494529724121,
|
|
"epoch": 0.01962929591968078,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.0189,
|
|
"mean_token_accuracy": 0.10438807159662247,
|
|
"num_tokens": 692816.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.758736324310303,
|
|
"epoch": 0.019951087656068994,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.1546,
|
|
"mean_token_accuracy": 0.10238918289542198,
|
|
"num_tokens": 705089.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.666284465789795,
|
|
"epoch": 0.020272879392457203,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.0252,
|
|
"mean_token_accuracy": 0.10788851976394653,
|
|
"num_tokens": 716874.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.531011438369751,
|
|
"epoch": 0.020594671128845412,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 6.9807,
|
|
"mean_token_accuracy": 0.10204816684126854,
|
|
"num_tokens": 728981.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.782111930847168,
|
|
"epoch": 0.02091646286523362,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.0871,
|
|
"mean_token_accuracy": 0.10574771091341972,
|
|
"num_tokens": 739619.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.599967575073242,
|
|
"epoch": 0.02123825460162183,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.0537,
|
|
"mean_token_accuracy": 0.10123220011591912,
|
|
"num_tokens": 751363.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.667736959457398,
|
|
"epoch": 0.02156004633801004,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.0722,
|
|
"mean_token_accuracy": 0.10424386709928513,
|
|
"num_tokens": 762953.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.489533710479736,
|
|
"epoch": 0.02188183807439825,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.9886,
|
|
"mean_token_accuracy": 0.10709020495414734,
|
|
"num_tokens": 775237.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.635052871704102,
|
|
"epoch": 0.022203629810786458,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.9133,
|
|
"mean_token_accuracy": 0.11272823959589004,
|
|
"num_tokens": 786262.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.458449077606201,
|
|
"epoch": 0.022525421547174667,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.9036,
|
|
"mean_token_accuracy": 0.11083822324872017,
|
|
"num_tokens": 797201.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.504961967468262,
|
|
"epoch": 0.02284721328356288,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.8593,
|
|
"mean_token_accuracy": 0.11371754407882691,
|
|
"num_tokens": 807671.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.4689586639404295,
|
|
"epoch": 0.02316900501995109,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.8925,
|
|
"mean_token_accuracy": 0.109658332914114,
|
|
"num_tokens": 819142.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.5210926055908205,
|
|
"epoch": 0.023490796756339298,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.8902,
|
|
"mean_token_accuracy": 0.11664389371871949,
|
|
"num_tokens": 829634.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.562618398666382,
|
|
"epoch": 0.023812588492727507,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.9457,
|
|
"mean_token_accuracy": 0.11507059112191201,
|
|
"num_tokens": 840723.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.44709038734436,
|
|
"epoch": 0.024134380229115716,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.9014,
|
|
"mean_token_accuracy": 0.11120686307549477,
|
|
"num_tokens": 852299.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.542909240722656,
|
|
"epoch": 0.024456171965503926,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.8845,
|
|
"mean_token_accuracy": 0.11418962627649307,
|
|
"num_tokens": 864427.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.269602823257446,
|
|
"epoch": 0.024777963701892135,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.8457,
|
|
"mean_token_accuracy": 0.11314126327633858,
|
|
"num_tokens": 876188.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.539252614974975,
|
|
"epoch": 0.025099755438280344,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.9751,
|
|
"mean_token_accuracy": 0.10968251451849938,
|
|
"num_tokens": 886835.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.462894248962402,
|
|
"epoch": 0.025421547174668553,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.8758,
|
|
"mean_token_accuracy": 0.1175595462322235,
|
|
"num_tokens": 899609.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.514445400238037,
|
|
"epoch": 0.025743338911056766,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.9867,
|
|
"mean_token_accuracy": 0.10821878015995026,
|
|
"num_tokens": 912114.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.434127473831177,
|
|
"epoch": 0.026065130647444975,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.9875,
|
|
"mean_token_accuracy": 0.1139479123055935,
|
|
"num_tokens": 923788.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.338033294677734,
|
|
"epoch": 0.026386922383833184,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.6791,
|
|
"mean_token_accuracy": 0.1255500763654709,
|
|
"num_tokens": 934952.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.422152519226074,
|
|
"epoch": 0.026708714120221393,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.7758,
|
|
"mean_token_accuracy": 0.11906903460621834,
|
|
"num_tokens": 945780.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.29374885559082,
|
|
"epoch": 0.027030505856609603,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.7765,
|
|
"mean_token_accuracy": 0.12052410319447518,
|
|
"num_tokens": 958502.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.422839784622193,
|
|
"epoch": 0.02735229759299781,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.8635,
|
|
"mean_token_accuracy": 0.11772049516439438,
|
|
"num_tokens": 969339.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.215829658508301,
|
|
"epoch": 0.02767408932938602,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.7171,
|
|
"mean_token_accuracy": 0.12233989164233208,
|
|
"num_tokens": 980260.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.2986561298370365,
|
|
"epoch": 0.02799588106577423,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.6963,
|
|
"mean_token_accuracy": 0.11892440766096116,
|
|
"num_tokens": 991960.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.193790006637573,
|
|
"epoch": 0.02831767280216244,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.7273,
|
|
"mean_token_accuracy": 0.11868971958756447,
|
|
"num_tokens": 1002083.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.308983230590821,
|
|
"epoch": 0.02863946453855065,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.7375,
|
|
"mean_token_accuracy": 0.11723077073693275,
|
|
"num_tokens": 1012896.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.334416151046753,
|
|
"epoch": 0.02896125627493886,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.8189,
|
|
"mean_token_accuracy": 0.12101824656128883,
|
|
"num_tokens": 1024570.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.2581829071044925,
|
|
"epoch": 0.02928304801132707,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.6596,
|
|
"mean_token_accuracy": 0.1274547331035137,
|
|
"num_tokens": 1035934.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.214243841171265,
|
|
"epoch": 0.02960483974771528,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.8555,
|
|
"mean_token_accuracy": 0.11035279110074044,
|
|
"num_tokens": 1047557.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.284564542770386,
|
|
"epoch": 0.02992663148410349,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.8011,
|
|
"mean_token_accuracy": 0.11536358669400215,
|
|
"num_tokens": 1059387.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.255402135848999,
|
|
"epoch": 0.030248423220491698,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.8003,
|
|
"mean_token_accuracy": 0.11578696817159653,
|
|
"num_tokens": 1070756.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.274528503417969,
|
|
"epoch": 0.030570214956879907,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.7873,
|
|
"mean_token_accuracy": 0.1295149154961109,
|
|
"num_tokens": 1083147.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.171627616882324,
|
|
"epoch": 0.030892006693268116,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.6541,
|
|
"mean_token_accuracy": 0.12174521461129188,
|
|
"num_tokens": 1094300.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.192223072052002,
|
|
"epoch": 0.031213798429656325,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.7988,
|
|
"mean_token_accuracy": 0.11465005055069924,
|
|
"num_tokens": 1105991.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.2635111808776855,
|
|
"epoch": 0.031535590166044535,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.8278,
|
|
"mean_token_accuracy": 0.11537677273154259,
|
|
"num_tokens": 1118065.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.217528486251831,
|
|
"epoch": 0.031857381902432744,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.772,
|
|
"mean_token_accuracy": 0.12680590748786927,
|
|
"num_tokens": 1130521.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.099181175231934,
|
|
"epoch": 0.03217917363882095,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.7436,
|
|
"mean_token_accuracy": 0.11461131051182746,
|
|
"num_tokens": 1143241.0,
|
|
"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": 271712573521920.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|