1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.010999890001099988,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.742588710784911,
|
|
"epoch": 0.00010999890001099989,
|
|
"grad_norm": 5.15625,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.716,
|
|
"mean_token_accuracy": 0.0001270647975616157,
|
|
"num_tokens": 7960.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.74256772994995,
|
|
"epoch": 0.00021999780002199978,
|
|
"grad_norm": 5.3125,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.6886,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 15356.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.742601013183593,
|
|
"epoch": 0.00032999670003299966,
|
|
"grad_norm": 5.3125,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.6881,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 22887.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.742615509033204,
|
|
"epoch": 0.00043999560004399956,
|
|
"grad_norm": 5.25,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.667,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 31706.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.74256649017334,
|
|
"epoch": 0.0005499945000549994,
|
|
"grad_norm": 4.6875,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.561,
|
|
"mean_token_accuracy": 0.001669197203591466,
|
|
"num_tokens": 40130.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.742358589172364,
|
|
"epoch": 0.0006599934000659993,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.4399,
|
|
"mean_token_accuracy": 0.01409623809158802,
|
|
"num_tokens": 48745.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.741664695739747,
|
|
"epoch": 0.0007699923000769992,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.3453,
|
|
"mean_token_accuracy": 0.03302886690944433,
|
|
"num_tokens": 58411.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.739883613586425,
|
|
"epoch": 0.0008799912000879991,
|
|
"grad_norm": 2.828125,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.1844,
|
|
"mean_token_accuracy": 0.03846868835389614,
|
|
"num_tokens": 66944.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.73689432144165,
|
|
"epoch": 0.000989990100098999,
|
|
"grad_norm": 2.46875,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.0458,
|
|
"mean_token_accuracy": 0.037658361718058586,
|
|
"num_tokens": 74600.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.73443374633789,
|
|
"epoch": 0.0010999890001099988,
|
|
"grad_norm": 2.25,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.9575,
|
|
"mean_token_accuracy": 0.0407950758934021,
|
|
"num_tokens": 82355.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.732843971252441,
|
|
"epoch": 0.0012099879001209988,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.9434,
|
|
"mean_token_accuracy": 0.03838752061128616,
|
|
"num_tokens": 90949.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.731368350982667,
|
|
"epoch": 0.0013199868001319986,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.8171,
|
|
"mean_token_accuracy": 0.04284884221851826,
|
|
"num_tokens": 98932.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.729646301269531,
|
|
"epoch": 0.0014299857001429986,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.8099,
|
|
"mean_token_accuracy": 0.03923225477337837,
|
|
"num_tokens": 107729.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.727366924285889,
|
|
"epoch": 0.0015399846001539984,
|
|
"grad_norm": 2.09375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.708,
|
|
"mean_token_accuracy": 0.04532702900469303,
|
|
"num_tokens": 116586.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.72254981994629,
|
|
"epoch": 0.0016499835001649984,
|
|
"grad_norm": 1.953125,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.6728,
|
|
"mean_token_accuracy": 0.041691455990076065,
|
|
"num_tokens": 124571.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.717883586883545,
|
|
"epoch": 0.0017599824001759982,
|
|
"grad_norm": 2.015625,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.6183,
|
|
"mean_token_accuracy": 0.04004398137331009,
|
|
"num_tokens": 133009.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.712346267700195,
|
|
"epoch": 0.001869981300186998,
|
|
"grad_norm": 1.9453125,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.4946,
|
|
"mean_token_accuracy": 0.04619385786354542,
|
|
"num_tokens": 141158.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.703997802734374,
|
|
"epoch": 0.001979980200197998,
|
|
"grad_norm": 1.9453125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.457,
|
|
"mean_token_accuracy": 0.04077398981899023,
|
|
"num_tokens": 148882.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.695512294769287,
|
|
"epoch": 0.002089979100208998,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.3832,
|
|
"mean_token_accuracy": 0.04717682003974914,
|
|
"num_tokens": 156679.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.684437274932861,
|
|
"epoch": 0.0021999780002199976,
|
|
"grad_norm": 1.984375,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.3058,
|
|
"mean_token_accuracy": 0.054728015512228015,
|
|
"num_tokens": 164165.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.668795394897462,
|
|
"epoch": 0.002309976900230998,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.2081,
|
|
"mean_token_accuracy": 0.054862821474671364,
|
|
"num_tokens": 172516.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.642210674285888,
|
|
"epoch": 0.0024199758002419977,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.1061,
|
|
"mean_token_accuracy": 0.057355142757296564,
|
|
"num_tokens": 181663.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.599918937683105,
|
|
"epoch": 0.0025299747002529975,
|
|
"grad_norm": 1.6953125,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 9.0113,
|
|
"mean_token_accuracy": 0.05499455593526363,
|
|
"num_tokens": 189346.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.538443756103515,
|
|
"epoch": 0.0026399736002639972,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.9581,
|
|
"mean_token_accuracy": 0.053924449160695075,
|
|
"num_tokens": 197631.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.470052242279053,
|
|
"epoch": 0.002749972500274997,
|
|
"grad_norm": 1.7890625,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.7814,
|
|
"mean_token_accuracy": 0.05690723732113838,
|
|
"num_tokens": 206443.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.407544708251953,
|
|
"epoch": 0.0028599714002859973,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.6333,
|
|
"mean_token_accuracy": 0.05704616643488407,
|
|
"num_tokens": 213629.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.335641956329345,
|
|
"epoch": 0.002969970300296997,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.5802,
|
|
"mean_token_accuracy": 0.052982112020254137,
|
|
"num_tokens": 221871.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.261301612854004,
|
|
"epoch": 0.003079969200307997,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.4161,
|
|
"mean_token_accuracy": 0.05874265432357788,
|
|
"num_tokens": 229725.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.158667469024659,
|
|
"epoch": 0.0031899681003189967,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.3519,
|
|
"mean_token_accuracy": 0.051278235763311385,
|
|
"num_tokens": 237438.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.028474044799804,
|
|
"epoch": 0.003299967000329997,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.2477,
|
|
"mean_token_accuracy": 0.05628519617021084,
|
|
"num_tokens": 245377.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.891870784759522,
|
|
"epoch": 0.0034099659003409967,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.1637,
|
|
"mean_token_accuracy": 0.05568413175642491,
|
|
"num_tokens": 254308.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.717917823791504,
|
|
"epoch": 0.0035199648003519965,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 8.0338,
|
|
"mean_token_accuracy": 0.05682709850370884,
|
|
"num_tokens": 261964.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.528457832336425,
|
|
"epoch": 0.0036299637003629963,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.914,
|
|
"mean_token_accuracy": 0.06191065795719623,
|
|
"num_tokens": 270225.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.34966287612915,
|
|
"epoch": 0.003739962600373996,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.8071,
|
|
"mean_token_accuracy": 0.06364000327885151,
|
|
"num_tokens": 277628.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.116067695617676,
|
|
"epoch": 0.0038499615003849963,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.8892,
|
|
"mean_token_accuracy": 0.06504783779382706,
|
|
"num_tokens": 286126.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.969095611572266,
|
|
"epoch": 0.003959960400395996,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.6584,
|
|
"mean_token_accuracy": 0.07001171819865704,
|
|
"num_tokens": 293439.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.705634784698486,
|
|
"epoch": 0.004069959300406996,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.6541,
|
|
"mean_token_accuracy": 0.06727977618575096,
|
|
"num_tokens": 302336.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.634931087493896,
|
|
"epoch": 0.004179958200417996,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.5934,
|
|
"mean_token_accuracy": 0.07421371676027774,
|
|
"num_tokens": 309605.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.477768993377685,
|
|
"epoch": 0.0042899571004289955,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.6276,
|
|
"mean_token_accuracy": 0.07245887257158756,
|
|
"num_tokens": 317722.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.315353870391846,
|
|
"epoch": 0.004399956000439995,
|
|
"grad_norm": 0.828125,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.5195,
|
|
"mean_token_accuracy": 0.07663455680012703,
|
|
"num_tokens": 325221.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.336180686950684,
|
|
"epoch": 0.004509954900450995,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.5851,
|
|
"mean_token_accuracy": 0.07097833044826984,
|
|
"num_tokens": 333548.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.292567348480224,
|
|
"epoch": 0.004619953800461996,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.4793,
|
|
"mean_token_accuracy": 0.07740762755274773,
|
|
"num_tokens": 341064.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.152500343322753,
|
|
"epoch": 0.0047299527004729955,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.4127,
|
|
"mean_token_accuracy": 0.08172049894928932,
|
|
"num_tokens": 349166.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.116729068756104,
|
|
"epoch": 0.004839951600483995,
|
|
"grad_norm": 0.8828125,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.5042,
|
|
"mean_token_accuracy": 0.07257169187068939,
|
|
"num_tokens": 358483.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.159700107574462,
|
|
"epoch": 0.004949950500494995,
|
|
"grad_norm": 0.84375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.4419,
|
|
"mean_token_accuracy": 0.07967055886983872,
|
|
"num_tokens": 366738.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.029997873306275,
|
|
"epoch": 0.005059949400505995,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.4202,
|
|
"mean_token_accuracy": 0.0808800846338272,
|
|
"num_tokens": 374438.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.9685265064239506,
|
|
"epoch": 0.005169948300516995,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.3842,
|
|
"mean_token_accuracy": 0.08380271010100841,
|
|
"num_tokens": 382721.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.05060772895813,
|
|
"epoch": 0.0052799472005279945,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.4199,
|
|
"mean_token_accuracy": 0.07848902978003025,
|
|
"num_tokens": 391631.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.95581464767456,
|
|
"epoch": 0.005389946100538994,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.3533,
|
|
"mean_token_accuracy": 0.08365190178155898,
|
|
"num_tokens": 400938.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.948987722396851,
|
|
"epoch": 0.005499945000549994,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.4494,
|
|
"mean_token_accuracy": 0.07313866652548313,
|
|
"num_tokens": 409462.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.000141048431397,
|
|
"epoch": 0.005609943900560995,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.4286,
|
|
"mean_token_accuracy": 0.08507802188396454,
|
|
"num_tokens": 416854.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.915297651290894,
|
|
"epoch": 0.0057199428005719945,
|
|
"grad_norm": 0.875,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.3407,
|
|
"mean_token_accuracy": 0.08099425658583641,
|
|
"num_tokens": 425008.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.934020614624023,
|
|
"epoch": 0.005829941700582994,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.3829,
|
|
"mean_token_accuracy": 0.08303864076733589,
|
|
"num_tokens": 433724.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.909064388275146,
|
|
"epoch": 0.005939940600593994,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.4365,
|
|
"mean_token_accuracy": 0.08042166382074356,
|
|
"num_tokens": 442054.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.977837800979614,
|
|
"epoch": 0.006049939500604994,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.3363,
|
|
"mean_token_accuracy": 0.08537820801138878,
|
|
"num_tokens": 448870.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.80058536529541,
|
|
"epoch": 0.006159938400615994,
|
|
"grad_norm": 0.8359375,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.2552,
|
|
"mean_token_accuracy": 0.08316881284117698,
|
|
"num_tokens": 457319.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.864862632751465,
|
|
"epoch": 0.0062699373006269935,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.4042,
|
|
"mean_token_accuracy": 0.08279576301574706,
|
|
"num_tokens": 466429.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.798379802703858,
|
|
"epoch": 0.006379936200637993,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.2791,
|
|
"mean_token_accuracy": 0.08732446394860745,
|
|
"num_tokens": 473054.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.787840890884399,
|
|
"epoch": 0.006489935100648993,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.2326,
|
|
"mean_token_accuracy": 0.08887541145086289,
|
|
"num_tokens": 480863.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.832312202453613,
|
|
"epoch": 0.006599934000659994,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.3602,
|
|
"mean_token_accuracy": 0.08132178895175457,
|
|
"num_tokens": 489619.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.7437928199768065,
|
|
"epoch": 0.0067099329006709936,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.1865,
|
|
"mean_token_accuracy": 0.08334602639079094,
|
|
"num_tokens": 497064.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.750201225280762,
|
|
"epoch": 0.006819931800681993,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.3193,
|
|
"mean_token_accuracy": 0.08315419107675552,
|
|
"num_tokens": 505855.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.780394983291626,
|
|
"epoch": 0.006929930700692993,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.3313,
|
|
"mean_token_accuracy": 0.08150239288806915,
|
|
"num_tokens": 514225.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.763619375228882,
|
|
"epoch": 0.007039929600703993,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.2511,
|
|
"mean_token_accuracy": 0.08538282066583633,
|
|
"num_tokens": 522286.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.820181655883789,
|
|
"epoch": 0.007149928500714993,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.2702,
|
|
"mean_token_accuracy": 0.08775243163108826,
|
|
"num_tokens": 531181.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.696579027175903,
|
|
"epoch": 0.0072599274007259925,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.3432,
|
|
"mean_token_accuracy": 0.08057724609971047,
|
|
"num_tokens": 540046.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.789985227584839,
|
|
"epoch": 0.007369926300736992,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.2619,
|
|
"mean_token_accuracy": 0.09030049964785576,
|
|
"num_tokens": 547512.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.773284912109375,
|
|
"epoch": 0.007479925200747992,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.2888,
|
|
"mean_token_accuracy": 0.08339077830314637,
|
|
"num_tokens": 555114.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.729707527160644,
|
|
"epoch": 0.007589924100758993,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.2806,
|
|
"mean_token_accuracy": 0.08401556685566902,
|
|
"num_tokens": 564719.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.7727906703948975,
|
|
"epoch": 0.007699923000769993,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.3146,
|
|
"mean_token_accuracy": 0.0900725856423378,
|
|
"num_tokens": 572263.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.685553455352784,
|
|
"epoch": 0.007809921900780992,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.142,
|
|
"mean_token_accuracy": 0.0930953487753868,
|
|
"num_tokens": 580518.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.635157585144043,
|
|
"epoch": 0.007919920800791992,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.1932,
|
|
"mean_token_accuracy": 0.08706289604306221,
|
|
"num_tokens": 589765.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.7066559314727785,
|
|
"epoch": 0.008029919700802992,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.1301,
|
|
"mean_token_accuracy": 0.09168279469013214,
|
|
"num_tokens": 597874.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.611246538162232,
|
|
"epoch": 0.008139918600813992,
|
|
"grad_norm": 0.9140625,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.1449,
|
|
"mean_token_accuracy": 0.09348057955503464,
|
|
"num_tokens": 605258.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.690258312225342,
|
|
"epoch": 0.008249917500824992,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.1931,
|
|
"mean_token_accuracy": 0.09372605606913567,
|
|
"num_tokens": 614059.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.649729585647583,
|
|
"epoch": 0.008359916400835991,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.1946,
|
|
"mean_token_accuracy": 0.09337117671966552,
|
|
"num_tokens": 622509.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.653411340713501,
|
|
"epoch": 0.008469915300846991,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.158,
|
|
"mean_token_accuracy": 0.08959888443350791,
|
|
"num_tokens": 630524.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.737890720367432,
|
|
"epoch": 0.008579914200857991,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.2129,
|
|
"mean_token_accuracy": 0.0871644526720047,
|
|
"num_tokens": 639629.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.542888069152832,
|
|
"epoch": 0.00868991310086899,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.2234,
|
|
"mean_token_accuracy": 0.08562670946121216,
|
|
"num_tokens": 648281.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.627482509613037,
|
|
"epoch": 0.00879991200087999,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 7.0552,
|
|
"mean_token_accuracy": 0.09867042228579521,
|
|
"num_tokens": 655925.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.558054399490357,
|
|
"epoch": 0.00890991090089099,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.9813,
|
|
"mean_token_accuracy": 0.10270329564809799,
|
|
"num_tokens": 663195.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.550625419616699,
|
|
"epoch": 0.00901990980090199,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 7.1238,
|
|
"mean_token_accuracy": 0.09005979970097541,
|
|
"num_tokens": 670935.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.657779312133789,
|
|
"epoch": 0.009129908700912992,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 0.000207,
|
|
"loss": 7.1409,
|
|
"mean_token_accuracy": 0.0887695387005806,
|
|
"num_tokens": 679995.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.622185134887696,
|
|
"epoch": 0.009239907600923991,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.1121,
|
|
"mean_token_accuracy": 0.09529180601239204,
|
|
"num_tokens": 688511.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.5377504348754885,
|
|
"epoch": 0.009349906500934991,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.000212,
|
|
"loss": 7.1059,
|
|
"mean_token_accuracy": 0.0890214629471302,
|
|
"num_tokens": 696346.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.524806690216065,
|
|
"epoch": 0.009459905400945991,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 7.1156,
|
|
"mean_token_accuracy": 0.0881573773920536,
|
|
"num_tokens": 705295.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.582342672348022,
|
|
"epoch": 0.00956990430095699,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 7.178,
|
|
"mean_token_accuracy": 0.08894185498356819,
|
|
"num_tokens": 712631.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.577776622772217,
|
|
"epoch": 0.00967990320096799,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 7.074,
|
|
"mean_token_accuracy": 0.09749704748392105,
|
|
"num_tokens": 720914.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.577001094818115,
|
|
"epoch": 0.00978990210097899,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.000222,
|
|
"loss": 7.1052,
|
|
"mean_token_accuracy": 0.09508850947022437,
|
|
"num_tokens": 729552.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.553100204467773,
|
|
"epoch": 0.00989990100098999,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 7.1556,
|
|
"mean_token_accuracy": 0.08776540756225586,
|
|
"num_tokens": 738361.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.443556928634644,
|
|
"epoch": 0.01000989990100099,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 7.0465,
|
|
"mean_token_accuracy": 0.09551195949316024,
|
|
"num_tokens": 746743.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.552993440628052,
|
|
"epoch": 0.01011989880101199,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 7.0491,
|
|
"mean_token_accuracy": 0.09229668080806733,
|
|
"num_tokens": 754410.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.436384820938111,
|
|
"epoch": 0.01022989770102299,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 7.0258,
|
|
"mean_token_accuracy": 0.09589382633566856,
|
|
"num_tokens": 761878.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.436898612976075,
|
|
"epoch": 0.01033989660103399,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 7.0382,
|
|
"mean_token_accuracy": 0.09157871827483177,
|
|
"num_tokens": 770217.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.475868082046508,
|
|
"epoch": 0.01044989550104499,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.929,
|
|
"mean_token_accuracy": 0.09891963005065918,
|
|
"num_tokens": 777915.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.526403570175171,
|
|
"epoch": 0.010559894401055989,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 7.1404,
|
|
"mean_token_accuracy": 0.08506608307361603,
|
|
"num_tokens": 788183.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.426519823074341,
|
|
"epoch": 0.010669893301066989,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.000242,
|
|
"loss": 7.0941,
|
|
"mean_token_accuracy": 0.08858079239726066,
|
|
"num_tokens": 796512.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.55730710029602,
|
|
"epoch": 0.010779892201077989,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 7.0438,
|
|
"mean_token_accuracy": 0.09578804671764374,
|
|
"num_tokens": 804157.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.450081729888916,
|
|
"epoch": 0.010889891101088988,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.000247,
|
|
"loss": 7.0147,
|
|
"mean_token_accuracy": 0.08979214057326317,
|
|
"num_tokens": 812315.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.391484498977661,
|
|
"epoch": 0.010999890001099988,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.9637,
|
|
"mean_token_accuracy": 0.10093949586153031,
|
|
"num_tokens": 820819.0,
|
|
"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": 226389615575040.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|