2035 lines
54 KiB
JSON
2035 lines
54 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.021999780002199976,
|
|
"eval_steps": 500,
|
|
"global_step": 1000,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.742588710784911,
|
|
"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
|
|
},
|
|
{
|
|
"entropy": 7.33015661239624,
|
|
"epoch": 0.01110988890111099,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.000252,
|
|
"loss": 6.9958,
|
|
"mean_token_accuracy": 0.09530256241559983,
|
|
"num_tokens": 828879.0,
|
|
"step": 505
|
|
},
|
|
{
|
|
"entropy": 7.519259977340698,
|
|
"epoch": 0.01121988780112199,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.0002545,
|
|
"loss": 7.0333,
|
|
"mean_token_accuracy": 0.09200103357434272,
|
|
"num_tokens": 837009.0,
|
|
"step": 510
|
|
},
|
|
{
|
|
"entropy": 7.391841125488281,
|
|
"epoch": 0.01132988670113299,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.000257,
|
|
"loss": 6.9979,
|
|
"mean_token_accuracy": 0.09302705451846123,
|
|
"num_tokens": 845245.0,
|
|
"step": 515
|
|
},
|
|
{
|
|
"entropy": 7.415077018737793,
|
|
"epoch": 0.011439885601143989,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.0002595,
|
|
"loss": 7.0139,
|
|
"mean_token_accuracy": 0.0946787141263485,
|
|
"num_tokens": 853472.0,
|
|
"step": 520
|
|
},
|
|
{
|
|
"entropy": 7.435290908813476,
|
|
"epoch": 0.011549884501154989,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000262,
|
|
"loss": 6.929,
|
|
"mean_token_accuracy": 0.09464162513613701,
|
|
"num_tokens": 861198.0,
|
|
"step": 525
|
|
},
|
|
{
|
|
"entropy": 7.35849962234497,
|
|
"epoch": 0.011659883401165989,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.00026450000000000003,
|
|
"loss": 6.9376,
|
|
"mean_token_accuracy": 0.09947292059659958,
|
|
"num_tokens": 868662.0,
|
|
"step": 530
|
|
},
|
|
{
|
|
"entropy": 7.440025424957275,
|
|
"epoch": 0.011769882301176988,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.00026700000000000004,
|
|
"loss": 7.0825,
|
|
"mean_token_accuracy": 0.0885132260620594,
|
|
"num_tokens": 877422.0,
|
|
"step": 535
|
|
},
|
|
{
|
|
"entropy": 7.3911055564880375,
|
|
"epoch": 0.011879881201187988,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00026950000000000005,
|
|
"loss": 6.9064,
|
|
"mean_token_accuracy": 0.11081198826432229,
|
|
"num_tokens": 885354.0,
|
|
"step": 540
|
|
},
|
|
{
|
|
"entropy": 7.352157926559448,
|
|
"epoch": 0.011989880101198988,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.00027200000000000005,
|
|
"loss": 6.93,
|
|
"mean_token_accuracy": 0.10218020156025887,
|
|
"num_tokens": 893763.0,
|
|
"step": 545
|
|
},
|
|
{
|
|
"entropy": 7.2934167861938475,
|
|
"epoch": 0.012099879001209988,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.0002745,
|
|
"loss": 6.8421,
|
|
"mean_token_accuracy": 0.10116673707962036,
|
|
"num_tokens": 901569.0,
|
|
"step": 550
|
|
},
|
|
{
|
|
"entropy": 7.456224060058593,
|
|
"epoch": 0.012209877901220988,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.000277,
|
|
"loss": 7.0442,
|
|
"mean_token_accuracy": 0.09027824476361275,
|
|
"num_tokens": 911208.0,
|
|
"step": 555
|
|
},
|
|
{
|
|
"entropy": 7.346185398101807,
|
|
"epoch": 0.012319876801231987,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.0002795,
|
|
"loss": 6.8846,
|
|
"mean_token_accuracy": 0.09844527244567872,
|
|
"num_tokens": 918350.0,
|
|
"step": 560
|
|
},
|
|
{
|
|
"entropy": 7.259835433959961,
|
|
"epoch": 0.012429875701242987,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00028199999999999997,
|
|
"loss": 6.9607,
|
|
"mean_token_accuracy": 0.10032025203108788,
|
|
"num_tokens": 927208.0,
|
|
"step": 565
|
|
},
|
|
{
|
|
"entropy": 7.368558406829834,
|
|
"epoch": 0.012539874601253987,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.0002845,
|
|
"loss": 6.9002,
|
|
"mean_token_accuracy": 0.09994437843561173,
|
|
"num_tokens": 935505.0,
|
|
"step": 570
|
|
},
|
|
{
|
|
"entropy": 7.261385726928711,
|
|
"epoch": 0.012649873501264987,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000287,
|
|
"loss": 6.8709,
|
|
"mean_token_accuracy": 0.09998608455061912,
|
|
"num_tokens": 942988.0,
|
|
"step": 575
|
|
},
|
|
{
|
|
"entropy": 7.492376041412354,
|
|
"epoch": 0.012759872401275987,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.0002895,
|
|
"loss": 6.9497,
|
|
"mean_token_accuracy": 0.09844574779272079,
|
|
"num_tokens": 950521.0,
|
|
"step": 580
|
|
},
|
|
{
|
|
"entropy": 7.22981333732605,
|
|
"epoch": 0.012869871301286986,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000292,
|
|
"loss": 6.9043,
|
|
"mean_token_accuracy": 0.10161162242293358,
|
|
"num_tokens": 959986.0,
|
|
"step": 585
|
|
},
|
|
{
|
|
"entropy": 7.290076017379761,
|
|
"epoch": 0.012979870201297986,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0002945,
|
|
"loss": 6.944,
|
|
"mean_token_accuracy": 0.09527330994606018,
|
|
"num_tokens": 969570.0,
|
|
"step": 590
|
|
},
|
|
{
|
|
"entropy": 7.37798228263855,
|
|
"epoch": 0.013089869101308988,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.000297,
|
|
"loss": 6.9375,
|
|
"mean_token_accuracy": 0.09366724938154221,
|
|
"num_tokens": 976847.0,
|
|
"step": 595
|
|
},
|
|
{
|
|
"entropy": 7.256167221069336,
|
|
"epoch": 0.013199868001319988,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.0002995,
|
|
"loss": 6.9007,
|
|
"mean_token_accuracy": 0.10105090960860252,
|
|
"num_tokens": 985768.0,
|
|
"step": 600
|
|
},
|
|
{
|
|
"entropy": 7.27320408821106,
|
|
"epoch": 0.013309866901330987,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000302,
|
|
"loss": 6.7888,
|
|
"mean_token_accuracy": 0.10461585223674774,
|
|
"num_tokens": 994960.0,
|
|
"step": 605
|
|
},
|
|
{
|
|
"entropy": 7.219715309143067,
|
|
"epoch": 0.013419865801341987,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0003045,
|
|
"loss": 6.8343,
|
|
"mean_token_accuracy": 0.10299366191029549,
|
|
"num_tokens": 1003099.0,
|
|
"step": 610
|
|
},
|
|
{
|
|
"entropy": 7.1929553031921385,
|
|
"epoch": 0.013529864701352987,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.000307,
|
|
"loss": 6.9241,
|
|
"mean_token_accuracy": 0.09887680411338806,
|
|
"num_tokens": 1012063.0,
|
|
"step": 615
|
|
},
|
|
{
|
|
"entropy": 7.275044202804565,
|
|
"epoch": 0.013639863601363987,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.0003095,
|
|
"loss": 6.8847,
|
|
"mean_token_accuracy": 0.10039810538291931,
|
|
"num_tokens": 1021849.0,
|
|
"step": 620
|
|
},
|
|
{
|
|
"entropy": 7.238638639450073,
|
|
"epoch": 0.013749862501374987,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.000312,
|
|
"loss": 6.9179,
|
|
"mean_token_accuracy": 0.0970994122326374,
|
|
"num_tokens": 1030701.0,
|
|
"step": 625
|
|
},
|
|
{
|
|
"entropy": 7.200835847854615,
|
|
"epoch": 0.013859861401385986,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0003145,
|
|
"loss": 6.9004,
|
|
"mean_token_accuracy": 0.09462883099913597,
|
|
"num_tokens": 1039743.0,
|
|
"step": 630
|
|
},
|
|
{
|
|
"entropy": 7.315715074539185,
|
|
"epoch": 0.013969860301396986,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000317,
|
|
"loss": 6.8128,
|
|
"mean_token_accuracy": 0.09855028465390206,
|
|
"num_tokens": 1049414.0,
|
|
"step": 635
|
|
},
|
|
{
|
|
"entropy": 7.194390439987183,
|
|
"epoch": 0.014079859201407986,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.0003195,
|
|
"loss": 6.8684,
|
|
"mean_token_accuracy": 0.10529875233769417,
|
|
"num_tokens": 1057361.0,
|
|
"step": 640
|
|
},
|
|
{
|
|
"entropy": 7.196909809112549,
|
|
"epoch": 0.014189858101418986,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.000322,
|
|
"loss": 6.8602,
|
|
"mean_token_accuracy": 0.10507860705256462,
|
|
"num_tokens": 1066091.0,
|
|
"step": 645
|
|
},
|
|
{
|
|
"entropy": 7.262992811203003,
|
|
"epoch": 0.014299857001429985,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.00032450000000000003,
|
|
"loss": 6.8386,
|
|
"mean_token_accuracy": 0.09974428489804268,
|
|
"num_tokens": 1074280.0,
|
|
"step": 650
|
|
},
|
|
{
|
|
"entropy": 7.1830676078796385,
|
|
"epoch": 0.014409855901440985,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.00032700000000000003,
|
|
"loss": 6.8171,
|
|
"mean_token_accuracy": 0.10386214479804039,
|
|
"num_tokens": 1081514.0,
|
|
"step": 655
|
|
},
|
|
{
|
|
"entropy": 7.232041645050049,
|
|
"epoch": 0.014519854801451985,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00032950000000000004,
|
|
"loss": 6.7957,
|
|
"mean_token_accuracy": 0.10456574410200119,
|
|
"num_tokens": 1089349.0,
|
|
"step": 660
|
|
},
|
|
{
|
|
"entropy": 7.215787744522094,
|
|
"epoch": 0.014629853701462985,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.00033200000000000005,
|
|
"loss": 6.8602,
|
|
"mean_token_accuracy": 0.10869375541806221,
|
|
"num_tokens": 1097779.0,
|
|
"step": 665
|
|
},
|
|
{
|
|
"entropy": 7.227393722534179,
|
|
"epoch": 0.014739852601473985,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.00033450000000000005,
|
|
"loss": 6.7632,
|
|
"mean_token_accuracy": 0.1064918652176857,
|
|
"num_tokens": 1105491.0,
|
|
"step": 670
|
|
},
|
|
{
|
|
"entropy": 7.103424978256226,
|
|
"epoch": 0.014849851501484984,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000337,
|
|
"loss": 6.8695,
|
|
"mean_token_accuracy": 0.1040018305182457,
|
|
"num_tokens": 1114180.0,
|
|
"step": 675
|
|
},
|
|
{
|
|
"entropy": 7.2291175365448,
|
|
"epoch": 0.014959850401495984,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0003395,
|
|
"loss": 6.7513,
|
|
"mean_token_accuracy": 0.10797639787197114,
|
|
"num_tokens": 1121874.0,
|
|
"step": 680
|
|
},
|
|
{
|
|
"entropy": 7.172188997268677,
|
|
"epoch": 0.015069849301506986,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000342,
|
|
"loss": 6.7646,
|
|
"mean_token_accuracy": 0.11012047156691551,
|
|
"num_tokens": 1130912.0,
|
|
"step": 685
|
|
},
|
|
{
|
|
"entropy": 7.113718175888062,
|
|
"epoch": 0.015179848201517986,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.00034449999999999997,
|
|
"loss": 6.7806,
|
|
"mean_token_accuracy": 0.10175866931676865,
|
|
"num_tokens": 1139991.0,
|
|
"step": 690
|
|
},
|
|
{
|
|
"entropy": 7.1981007099151615,
|
|
"epoch": 0.015289847101528985,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.000347,
|
|
"loss": 6.8294,
|
|
"mean_token_accuracy": 0.10432724431157112,
|
|
"num_tokens": 1147884.0,
|
|
"step": 695
|
|
},
|
|
{
|
|
"entropy": 7.170119333267212,
|
|
"epoch": 0.015399846001539985,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.0003495,
|
|
"loss": 6.8714,
|
|
"mean_token_accuracy": 0.09817730337381363,
|
|
"num_tokens": 1155306.0,
|
|
"step": 700
|
|
},
|
|
{
|
|
"entropy": 7.228573799133301,
|
|
"epoch": 0.015509844901550985,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.000352,
|
|
"loss": 6.7851,
|
|
"mean_token_accuracy": 0.09527272805571556,
|
|
"num_tokens": 1163063.0,
|
|
"step": 705
|
|
},
|
|
{
|
|
"entropy": 7.073216915130615,
|
|
"epoch": 0.015619843801561985,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0003545,
|
|
"loss": 6.826,
|
|
"mean_token_accuracy": 0.09759803414344788,
|
|
"num_tokens": 1172766.0,
|
|
"step": 710
|
|
},
|
|
{
|
|
"entropy": 7.1333136558532715,
|
|
"epoch": 0.015729842701572985,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.000357,
|
|
"loss": 6.7568,
|
|
"mean_token_accuracy": 0.10859671384096145,
|
|
"num_tokens": 1180185.0,
|
|
"step": 715
|
|
},
|
|
{
|
|
"entropy": 7.180244493484497,
|
|
"epoch": 0.015839841601583984,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0003595,
|
|
"loss": 6.7884,
|
|
"mean_token_accuracy": 0.11051692590117454,
|
|
"num_tokens": 1188354.0,
|
|
"step": 720
|
|
},
|
|
{
|
|
"entropy": 7.020302581787109,
|
|
"epoch": 0.015949840501594984,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000362,
|
|
"loss": 6.7896,
|
|
"mean_token_accuracy": 0.10646956786513329,
|
|
"num_tokens": 1196887.0,
|
|
"step": 725
|
|
},
|
|
{
|
|
"entropy": 7.2585468769073485,
|
|
"epoch": 0.016059839401605984,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0003645,
|
|
"loss": 6.6886,
|
|
"mean_token_accuracy": 0.11256491467356682,
|
|
"num_tokens": 1204577.0,
|
|
"step": 730
|
|
},
|
|
{
|
|
"entropy": 7.026610612869263,
|
|
"epoch": 0.016169838301616984,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000367,
|
|
"loss": 6.6218,
|
|
"mean_token_accuracy": 0.11297912001609803,
|
|
"num_tokens": 1211548.0,
|
|
"step": 735
|
|
},
|
|
{
|
|
"entropy": 7.19339952468872,
|
|
"epoch": 0.016279837201627984,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.0003695,
|
|
"loss": 6.7862,
|
|
"mean_token_accuracy": 0.10827388316392898,
|
|
"num_tokens": 1218770.0,
|
|
"step": 740
|
|
},
|
|
{
|
|
"entropy": 7.0089672088623045,
|
|
"epoch": 0.016389836101638983,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.000372,
|
|
"loss": 6.605,
|
|
"mean_token_accuracy": 0.12037925198674201,
|
|
"num_tokens": 1226268.0,
|
|
"step": 745
|
|
},
|
|
{
|
|
"entropy": 7.073901653289795,
|
|
"epoch": 0.016499835001649983,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0003745,
|
|
"loss": 6.7904,
|
|
"mean_token_accuracy": 0.10318765193223953,
|
|
"num_tokens": 1233629.0,
|
|
"step": 750
|
|
},
|
|
{
|
|
"entropy": 7.133617353439331,
|
|
"epoch": 0.016609833901660983,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.000377,
|
|
"loss": 6.8259,
|
|
"mean_token_accuracy": 0.10541872754693031,
|
|
"num_tokens": 1240892.0,
|
|
"step": 755
|
|
},
|
|
{
|
|
"entropy": 7.0288292407989506,
|
|
"epoch": 0.016719832801671983,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0003795,
|
|
"loss": 6.6407,
|
|
"mean_token_accuracy": 0.10865572765469551,
|
|
"num_tokens": 1247898.0,
|
|
"step": 760
|
|
},
|
|
{
|
|
"entropy": 7.055471658706665,
|
|
"epoch": 0.016829831701682983,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000382,
|
|
"loss": 6.7682,
|
|
"mean_token_accuracy": 0.10503391325473785,
|
|
"num_tokens": 1257066.0,
|
|
"step": 765
|
|
},
|
|
{
|
|
"entropy": 7.13066349029541,
|
|
"epoch": 0.016939830601693982,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0003845,
|
|
"loss": 6.8114,
|
|
"mean_token_accuracy": 0.1041012242436409,
|
|
"num_tokens": 1266225.0,
|
|
"step": 770
|
|
},
|
|
{
|
|
"entropy": 7.004931020736694,
|
|
"epoch": 0.017049829501704982,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.00038700000000000003,
|
|
"loss": 6.7009,
|
|
"mean_token_accuracy": 0.11033282876014709,
|
|
"num_tokens": 1274318.0,
|
|
"step": 775
|
|
},
|
|
{
|
|
"entropy": 7.223189496994019,
|
|
"epoch": 0.017159828401715982,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.00038950000000000003,
|
|
"loss": 6.8503,
|
|
"mean_token_accuracy": 0.10478579625487328,
|
|
"num_tokens": 1282636.0,
|
|
"step": 780
|
|
},
|
|
{
|
|
"entropy": 7.068056297302246,
|
|
"epoch": 0.01726982730172698,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.00039200000000000004,
|
|
"loss": 6.6899,
|
|
"mean_token_accuracy": 0.11055590659379959,
|
|
"num_tokens": 1290479.0,
|
|
"step": 785
|
|
},
|
|
{
|
|
"entropy": 7.065642690658569,
|
|
"epoch": 0.01737982620173798,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00039450000000000005,
|
|
"loss": 6.7652,
|
|
"mean_token_accuracy": 0.10867261067032814,
|
|
"num_tokens": 1299907.0,
|
|
"step": 790
|
|
},
|
|
{
|
|
"entropy": 7.026779890060425,
|
|
"epoch": 0.01748982510174898,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.00039700000000000005,
|
|
"loss": 6.6163,
|
|
"mean_token_accuracy": 0.12043424397706985,
|
|
"num_tokens": 1307204.0,
|
|
"step": 795
|
|
},
|
|
{
|
|
"entropy": 7.105731725692749,
|
|
"epoch": 0.01759982400175998,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0003995,
|
|
"loss": 6.798,
|
|
"mean_token_accuracy": 0.11027551516890526,
|
|
"num_tokens": 1315779.0,
|
|
"step": 800
|
|
},
|
|
{
|
|
"entropy": 7.073409748077393,
|
|
"epoch": 0.01770982290177098,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.000402,
|
|
"loss": 6.755,
|
|
"mean_token_accuracy": 0.1137322336435318,
|
|
"num_tokens": 1323270.0,
|
|
"step": 805
|
|
},
|
|
{
|
|
"entropy": 7.003409004211425,
|
|
"epoch": 0.01781982180178198,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.0004045,
|
|
"loss": 6.6495,
|
|
"mean_token_accuracy": 0.108744066208601,
|
|
"num_tokens": 1331955.0,
|
|
"step": 810
|
|
},
|
|
{
|
|
"entropy": 7.022155809402466,
|
|
"epoch": 0.01792982070179298,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.00040699999999999997,
|
|
"loss": 6.6892,
|
|
"mean_token_accuracy": 0.10952874571084976,
|
|
"num_tokens": 1339432.0,
|
|
"step": 815
|
|
},
|
|
{
|
|
"entropy": 6.950418281555176,
|
|
"epoch": 0.01803981960180398,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.0004095,
|
|
"loss": 6.6766,
|
|
"mean_token_accuracy": 0.10625928342342376,
|
|
"num_tokens": 1347269.0,
|
|
"step": 820
|
|
},
|
|
{
|
|
"entropy": 7.131736421585083,
|
|
"epoch": 0.018149818501814984,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.000412,
|
|
"loss": 6.6822,
|
|
"mean_token_accuracy": 0.11529266983270645,
|
|
"num_tokens": 1356373.0,
|
|
"step": 825
|
|
},
|
|
{
|
|
"entropy": 6.912766456604004,
|
|
"epoch": 0.018259817401825983,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0004145,
|
|
"loss": 6.7527,
|
|
"mean_token_accuracy": 0.10752285644412041,
|
|
"num_tokens": 1365650.0,
|
|
"step": 830
|
|
},
|
|
{
|
|
"entropy": 7.066224718093872,
|
|
"epoch": 0.018369816301836983,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.000417,
|
|
"loss": 6.6481,
|
|
"mean_token_accuracy": 0.1153008408844471,
|
|
"num_tokens": 1372903.0,
|
|
"step": 835
|
|
},
|
|
{
|
|
"entropy": 7.01744818687439,
|
|
"epoch": 0.018479815201847983,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.0004195,
|
|
"loss": 6.6401,
|
|
"mean_token_accuracy": 0.11252194419503211,
|
|
"num_tokens": 1380789.0,
|
|
"step": 840
|
|
},
|
|
{
|
|
"entropy": 7.041985893249512,
|
|
"epoch": 0.018589814101858983,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.000422,
|
|
"loss": 6.7752,
|
|
"mean_token_accuracy": 0.1045458309352398,
|
|
"num_tokens": 1389980.0,
|
|
"step": 845
|
|
},
|
|
{
|
|
"entropy": 6.92493371963501,
|
|
"epoch": 0.018699813001869982,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0004245,
|
|
"loss": 6.681,
|
|
"mean_token_accuracy": 0.10559192821383476,
|
|
"num_tokens": 1397786.0,
|
|
"step": 850
|
|
},
|
|
{
|
|
"entropy": 7.1217365741729735,
|
|
"epoch": 0.018809811901880982,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.000427,
|
|
"loss": 6.7299,
|
|
"mean_token_accuracy": 0.1081755816936493,
|
|
"num_tokens": 1406063.0,
|
|
"step": 855
|
|
},
|
|
{
|
|
"entropy": 7.054189825057984,
|
|
"epoch": 0.018919810801891982,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.0004295,
|
|
"loss": 6.6997,
|
|
"mean_token_accuracy": 0.10684741958975792,
|
|
"num_tokens": 1415669.0,
|
|
"step": 860
|
|
},
|
|
{
|
|
"entropy": 6.798714971542358,
|
|
"epoch": 0.019029809701902982,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.000432,
|
|
"loss": 6.5361,
|
|
"mean_token_accuracy": 0.11739783585071564,
|
|
"num_tokens": 1424084.0,
|
|
"step": 865
|
|
},
|
|
{
|
|
"entropy": 7.0029017448425295,
|
|
"epoch": 0.01913980860191398,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.0004345,
|
|
"loss": 6.7152,
|
|
"mean_token_accuracy": 0.11012220978736878,
|
|
"num_tokens": 1433489.0,
|
|
"step": 870
|
|
},
|
|
{
|
|
"entropy": 6.969244861602784,
|
|
"epoch": 0.01924980750192498,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.000437,
|
|
"loss": 6.589,
|
|
"mean_token_accuracy": 0.11142838001251221,
|
|
"num_tokens": 1441630.0,
|
|
"step": 875
|
|
},
|
|
{
|
|
"entropy": 6.89876880645752,
|
|
"epoch": 0.01935980640193598,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0004395,
|
|
"loss": 6.6666,
|
|
"mean_token_accuracy": 0.10799175277352333,
|
|
"num_tokens": 1449362.0,
|
|
"step": 880
|
|
},
|
|
{
|
|
"entropy": 6.860118198394775,
|
|
"epoch": 0.01946980530194698,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.000442,
|
|
"loss": 6.606,
|
|
"mean_token_accuracy": 0.10998807772994042,
|
|
"num_tokens": 1456087.0,
|
|
"step": 885
|
|
},
|
|
{
|
|
"entropy": 7.058117389678955,
|
|
"epoch": 0.01957980420195798,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.0004445,
|
|
"loss": 6.7236,
|
|
"mean_token_accuracy": 0.10595543459057807,
|
|
"num_tokens": 1463159.0,
|
|
"step": 890
|
|
},
|
|
{
|
|
"entropy": 7.008705186843872,
|
|
"epoch": 0.01968980310196898,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000447,
|
|
"loss": 6.7106,
|
|
"mean_token_accuracy": 0.10578989833593369,
|
|
"num_tokens": 1472228.0,
|
|
"step": 895
|
|
},
|
|
{
|
|
"entropy": 6.925995206832885,
|
|
"epoch": 0.01979980200197998,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00044950000000000003,
|
|
"loss": 6.5438,
|
|
"mean_token_accuracy": 0.11546369567513466,
|
|
"num_tokens": 1479634.0,
|
|
"step": 900
|
|
},
|
|
{
|
|
"entropy": 6.857554626464844,
|
|
"epoch": 0.01990980090199098,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00045200000000000004,
|
|
"loss": 6.6233,
|
|
"mean_token_accuracy": 0.11085652485489846,
|
|
"num_tokens": 1488295.0,
|
|
"step": 905
|
|
},
|
|
{
|
|
"entropy": 6.982423162460327,
|
|
"epoch": 0.02001979980200198,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00045450000000000004,
|
|
"loss": 6.6722,
|
|
"mean_token_accuracy": 0.1068225473165512,
|
|
"num_tokens": 1495356.0,
|
|
"step": 910
|
|
},
|
|
{
|
|
"entropy": 6.915511560440064,
|
|
"epoch": 0.02012979870201298,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.00045700000000000005,
|
|
"loss": 6.6213,
|
|
"mean_token_accuracy": 0.11951702162623405,
|
|
"num_tokens": 1502813.0,
|
|
"step": 915
|
|
},
|
|
{
|
|
"entropy": 6.919120836257934,
|
|
"epoch": 0.02023979760202398,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00045950000000000006,
|
|
"loss": 6.5949,
|
|
"mean_token_accuracy": 0.10837950855493546,
|
|
"num_tokens": 1510325.0,
|
|
"step": 920
|
|
},
|
|
{
|
|
"entropy": 6.888558721542358,
|
|
"epoch": 0.02034979650203498,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.000462,
|
|
"loss": 6.5728,
|
|
"mean_token_accuracy": 0.11481452733278275,
|
|
"num_tokens": 1518457.0,
|
|
"step": 925
|
|
},
|
|
{
|
|
"entropy": 6.859040880203247,
|
|
"epoch": 0.02045979540204598,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 0.0004645,
|
|
"loss": 6.5773,
|
|
"mean_token_accuracy": 0.1093935489654541,
|
|
"num_tokens": 1527007.0,
|
|
"step": 930
|
|
},
|
|
{
|
|
"entropy": 6.907317495346069,
|
|
"epoch": 0.02056979430205698,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.000467,
|
|
"loss": 6.4604,
|
|
"mean_token_accuracy": 0.11935687363147736,
|
|
"num_tokens": 1534636.0,
|
|
"step": 935
|
|
},
|
|
{
|
|
"entropy": 6.78266339302063,
|
|
"epoch": 0.02067979320206798,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.0004695,
|
|
"loss": 6.5993,
|
|
"mean_token_accuracy": 0.11558415740728378,
|
|
"num_tokens": 1542302.0,
|
|
"step": 940
|
|
},
|
|
{
|
|
"entropy": 6.860093975067139,
|
|
"epoch": 0.02078979210207898,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000472,
|
|
"loss": 6.6071,
|
|
"mean_token_accuracy": 0.11185053288936615,
|
|
"num_tokens": 1550649.0,
|
|
"step": 945
|
|
},
|
|
{
|
|
"entropy": 6.824812364578247,
|
|
"epoch": 0.02089979100208998,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0004745,
|
|
"loss": 6.5235,
|
|
"mean_token_accuracy": 0.11756469607353211,
|
|
"num_tokens": 1558392.0,
|
|
"step": 950
|
|
},
|
|
{
|
|
"entropy": 6.851763677597046,
|
|
"epoch": 0.021009789902100978,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000477,
|
|
"loss": 6.5286,
|
|
"mean_token_accuracy": 0.1210299015045166,
|
|
"num_tokens": 1566249.0,
|
|
"step": 955
|
|
},
|
|
{
|
|
"entropy": 6.849990558624268,
|
|
"epoch": 0.021119788802111978,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0004795,
|
|
"loss": 6.6454,
|
|
"mean_token_accuracy": 0.1076412208378315,
|
|
"num_tokens": 1576055.0,
|
|
"step": 960
|
|
},
|
|
{
|
|
"entropy": 6.788512277603149,
|
|
"epoch": 0.021229787702122978,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000482,
|
|
"loss": 6.5131,
|
|
"mean_token_accuracy": 0.11613129451870918,
|
|
"num_tokens": 1584590.0,
|
|
"step": 965
|
|
},
|
|
{
|
|
"entropy": 6.838148069381714,
|
|
"epoch": 0.021339786602133978,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.0004845,
|
|
"loss": 6.5421,
|
|
"mean_token_accuracy": 0.11711058914661407,
|
|
"num_tokens": 1593085.0,
|
|
"step": 970
|
|
},
|
|
{
|
|
"entropy": 6.903753995895386,
|
|
"epoch": 0.021449785502144977,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000487,
|
|
"loss": 6.5696,
|
|
"mean_token_accuracy": 0.11010838970541954,
|
|
"num_tokens": 1601181.0,
|
|
"step": 975
|
|
},
|
|
{
|
|
"entropy": 6.699677991867065,
|
|
"epoch": 0.021559784402155977,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.0004895,
|
|
"loss": 6.4763,
|
|
"mean_token_accuracy": 0.12405971437692642,
|
|
"num_tokens": 1608857.0,
|
|
"step": 980
|
|
},
|
|
{
|
|
"entropy": 6.864647722244262,
|
|
"epoch": 0.021669783302166977,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000492,
|
|
"loss": 6.5545,
|
|
"mean_token_accuracy": 0.11281617730855942,
|
|
"num_tokens": 1617592.0,
|
|
"step": 985
|
|
},
|
|
{
|
|
"entropy": 6.803263378143311,
|
|
"epoch": 0.021779782202177977,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0004945,
|
|
"loss": 6.576,
|
|
"mean_token_accuracy": 0.10787124335765838,
|
|
"num_tokens": 1626383.0,
|
|
"step": 990
|
|
},
|
|
{
|
|
"entropy": 6.861760568618775,
|
|
"epoch": 0.021889781102188977,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.000497,
|
|
"loss": 6.5253,
|
|
"mean_token_accuracy": 0.12360839247703552,
|
|
"num_tokens": 1634806.0,
|
|
"step": 995
|
|
},
|
|
{
|
|
"entropy": 6.699004888534546,
|
|
"epoch": 0.021999780002199976,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0004995,
|
|
"loss": 6.5902,
|
|
"mean_token_accuracy": 0.11484271436929702,
|
|
"num_tokens": 1642897.0,
|
|
"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": 460964190781440.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|