1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.01353637901861252,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.69194974899292,
|
|
"epoch": 0.00013536379018612522,
|
|
"grad_norm": 12.5,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.8777,
|
|
"mean_token_accuracy": 5.577245028689504e-05,
|
|
"num_tokens": 15054.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691948223114014,
|
|
"epoch": 0.00027072758037225043,
|
|
"grad_norm": 12.0625,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.8153,
|
|
"mean_token_accuracy": 0.00044745010090991856,
|
|
"num_tokens": 28250.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.691227531433105,
|
|
"epoch": 0.0004060913705583756,
|
|
"grad_norm": 9.1875,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.5817,
|
|
"mean_token_accuracy": 0.0174422716954723,
|
|
"num_tokens": 41527.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.685559177398682,
|
|
"epoch": 0.0005414551607445009,
|
|
"grad_norm": 5.9375,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.2793,
|
|
"mean_token_accuracy": 0.034390256740152836,
|
|
"num_tokens": 54285.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.661216259002686,
|
|
"epoch": 0.0006768189509306261,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.0155,
|
|
"mean_token_accuracy": 0.031836986914277075,
|
|
"num_tokens": 69833.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.635576629638672,
|
|
"epoch": 0.0008121827411167512,
|
|
"grad_norm": 2.9375,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.8988,
|
|
"mean_token_accuracy": 0.033391524106264114,
|
|
"num_tokens": 84065.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.61604061126709,
|
|
"epoch": 0.0009475465313028764,
|
|
"grad_norm": 2.375,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.7824,
|
|
"mean_token_accuracy": 0.035429037548601626,
|
|
"num_tokens": 97347.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.616745281219483,
|
|
"epoch": 0.0010829103214890017,
|
|
"grad_norm": 2.5,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.722,
|
|
"mean_token_accuracy": 0.034273325465619564,
|
|
"num_tokens": 110883.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.620021438598632,
|
|
"epoch": 0.001218274111675127,
|
|
"grad_norm": 2.1875,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.7299,
|
|
"mean_token_accuracy": 0.031383275240659717,
|
|
"num_tokens": 125407.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.61033935546875,
|
|
"epoch": 0.0013536379018612521,
|
|
"grad_norm": 2.390625,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.6781,
|
|
"mean_token_accuracy": 0.03162609338760376,
|
|
"num_tokens": 138221.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.609558010101319,
|
|
"epoch": 0.0014890016920473773,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.6075,
|
|
"mean_token_accuracy": 0.037006295286118984,
|
|
"num_tokens": 150716.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.592256355285645,
|
|
"epoch": 0.0016243654822335025,
|
|
"grad_norm": 2.15625,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.5363,
|
|
"mean_token_accuracy": 0.033432438410818575,
|
|
"num_tokens": 167084.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.596331214904785,
|
|
"epoch": 0.0017597292724196277,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.5357,
|
|
"mean_token_accuracy": 0.028936176747083663,
|
|
"num_tokens": 181029.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.597533226013184,
|
|
"epoch": 0.0018950930626057529,
|
|
"grad_norm": 2.09375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.3621,
|
|
"mean_token_accuracy": 0.03304864317178726,
|
|
"num_tokens": 194629.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.554742336273193,
|
|
"epoch": 0.0020304568527918783,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.2827,
|
|
"mean_token_accuracy": 0.029723377712070943,
|
|
"num_tokens": 207367.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.533198165893555,
|
|
"epoch": 0.0021658206429780035,
|
|
"grad_norm": 1.9453125,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.1578,
|
|
"mean_token_accuracy": 0.029196519777178763,
|
|
"num_tokens": 221208.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.477063274383545,
|
|
"epoch": 0.0023011844331641287,
|
|
"grad_norm": 2.015625,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.0349,
|
|
"mean_token_accuracy": 0.03920750468969345,
|
|
"num_tokens": 233176.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.465512084960938,
|
|
"epoch": 0.002436548223350254,
|
|
"grad_norm": 2.015625,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.9479,
|
|
"mean_token_accuracy": 0.037744084000587465,
|
|
"num_tokens": 247745.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.410159492492676,
|
|
"epoch": 0.002571912013536379,
|
|
"grad_norm": 2.578125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.8433,
|
|
"mean_token_accuracy": 0.037843816168606284,
|
|
"num_tokens": 262129.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.357386016845703,
|
|
"epoch": 0.0027072758037225042,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.7074,
|
|
"mean_token_accuracy": 0.042482448555529115,
|
|
"num_tokens": 275500.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.285986518859863,
|
|
"epoch": 0.0028426395939086294,
|
|
"grad_norm": 1.875,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.6332,
|
|
"mean_token_accuracy": 0.04186442382633686,
|
|
"num_tokens": 290437.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.227664184570312,
|
|
"epoch": 0.0029780033840947546,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.4353,
|
|
"mean_token_accuracy": 0.0504638671875,
|
|
"num_tokens": 304615.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.08832664489746,
|
|
"epoch": 0.00311336717428088,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.263,
|
|
"mean_token_accuracy": 0.05822908580303192,
|
|
"num_tokens": 318560.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.972474098205566,
|
|
"epoch": 0.003248730964467005,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.1931,
|
|
"mean_token_accuracy": 0.05676005966961384,
|
|
"num_tokens": 332748.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.847494888305665,
|
|
"epoch": 0.00338409475465313,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 7.9947,
|
|
"mean_token_accuracy": 0.06220810636878014,
|
|
"num_tokens": 347127.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.678038311004638,
|
|
"epoch": 0.0035194585448392554,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 7.9391,
|
|
"mean_token_accuracy": 0.058373385295271876,
|
|
"num_tokens": 359407.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.540247344970703,
|
|
"epoch": 0.0036548223350253805,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 7.7636,
|
|
"mean_token_accuracy": 0.061562583222985266,
|
|
"num_tokens": 375021.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.30407886505127,
|
|
"epoch": 0.0037901861252115057,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.5962,
|
|
"mean_token_accuracy": 0.070381173864007,
|
|
"num_tokens": 389496.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.132857418060302,
|
|
"epoch": 0.003925549915397631,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.6112,
|
|
"mean_token_accuracy": 0.06686735302209854,
|
|
"num_tokens": 402465.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.873841571807862,
|
|
"epoch": 0.0040609137055837565,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.502,
|
|
"mean_token_accuracy": 0.0693404596298933,
|
|
"num_tokens": 415980.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.732354259490966,
|
|
"epoch": 0.004196277495769882,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.4634,
|
|
"mean_token_accuracy": 0.07001067139208317,
|
|
"num_tokens": 428623.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.544918155670166,
|
|
"epoch": 0.004331641285956007,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.4298,
|
|
"mean_token_accuracy": 0.06541897803544998,
|
|
"num_tokens": 441999.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.378799533843994,
|
|
"epoch": 0.004467005076142132,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.3005,
|
|
"mean_token_accuracy": 0.07170183844864368,
|
|
"num_tokens": 455098.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.280582523345947,
|
|
"epoch": 0.004602368866328257,
|
|
"grad_norm": 1.9375,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.2898,
|
|
"mean_token_accuracy": 0.07008971199393273,
|
|
"num_tokens": 468506.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.127180671691894,
|
|
"epoch": 0.0047377326565143825,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.3215,
|
|
"mean_token_accuracy": 0.06721526682376862,
|
|
"num_tokens": 482795.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.071395254135131,
|
|
"epoch": 0.004873096446700508,
|
|
"grad_norm": 1.7109375,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.2161,
|
|
"mean_token_accuracy": 0.07408471405506134,
|
|
"num_tokens": 497102.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.015200138092041,
|
|
"epoch": 0.005008460236886633,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.236,
|
|
"mean_token_accuracy": 0.07463390305638314,
|
|
"num_tokens": 511777.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.013217878341674,
|
|
"epoch": 0.005143824027072758,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.2509,
|
|
"mean_token_accuracy": 0.07550472915172576,
|
|
"num_tokens": 525855.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 7.909638738632202,
|
|
"epoch": 0.005279187817258883,
|
|
"grad_norm": 1.84375,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.1878,
|
|
"mean_token_accuracy": 0.0701899241656065,
|
|
"num_tokens": 541364.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.90043454170227,
|
|
"epoch": 0.0054145516074450084,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.1873,
|
|
"mean_token_accuracy": 0.0761424608528614,
|
|
"num_tokens": 555489.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.926745128631592,
|
|
"epoch": 0.005549915397631134,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.2169,
|
|
"mean_token_accuracy": 0.07424422949552537,
|
|
"num_tokens": 569207.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.8149010181427006,
|
|
"epoch": 0.005685279187817259,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.1699,
|
|
"mean_token_accuracy": 0.07592894844710826,
|
|
"num_tokens": 583689.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.787826442718506,
|
|
"epoch": 0.005820642978003384,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.1253,
|
|
"mean_token_accuracy": 0.0805364765226841,
|
|
"num_tokens": 597729.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.7665050506591795,
|
|
"epoch": 0.005956006768189509,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.1149,
|
|
"mean_token_accuracy": 0.08125128298997879,
|
|
"num_tokens": 610354.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.738793563842774,
|
|
"epoch": 0.006091370558375634,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 6.9965,
|
|
"mean_token_accuracy": 0.08436518236994743,
|
|
"num_tokens": 625591.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.677218818664551,
|
|
"epoch": 0.00622673434856176,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.1185,
|
|
"mean_token_accuracy": 0.08086443841457366,
|
|
"num_tokens": 640936.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.736022233963013,
|
|
"epoch": 0.006362098138747885,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.0841,
|
|
"mean_token_accuracy": 0.0765869714319706,
|
|
"num_tokens": 655949.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.609257936477661,
|
|
"epoch": 0.00649746192893401,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 6.948,
|
|
"mean_token_accuracy": 0.08549271896481514,
|
|
"num_tokens": 669357.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.508241319656372,
|
|
"epoch": 0.006632825719120135,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.000122,
|
|
"loss": 6.8378,
|
|
"mean_token_accuracy": 0.08860795795917512,
|
|
"num_tokens": 683906.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.5739704132080075,
|
|
"epoch": 0.00676818950930626,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 6.9511,
|
|
"mean_token_accuracy": 0.0863402545452118,
|
|
"num_tokens": 700467.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.5351649761199955,
|
|
"epoch": 0.0069035532994923855,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.000127,
|
|
"loss": 6.9582,
|
|
"mean_token_accuracy": 0.09148906618356704,
|
|
"num_tokens": 711779.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.530111169815063,
|
|
"epoch": 0.007038917089678511,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 6.9595,
|
|
"mean_token_accuracy": 0.08548436611890793,
|
|
"num_tokens": 725455.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.483809947967529,
|
|
"epoch": 0.007174280879864636,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.000132,
|
|
"loss": 6.8334,
|
|
"mean_token_accuracy": 0.08580193966627121,
|
|
"num_tokens": 740428.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.460272359848022,
|
|
"epoch": 0.007309644670050761,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 6.8215,
|
|
"mean_token_accuracy": 0.09169955775141717,
|
|
"num_tokens": 755327.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.53057975769043,
|
|
"epoch": 0.007445008460236886,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 6.9262,
|
|
"mean_token_accuracy": 0.09086679890751839,
|
|
"num_tokens": 768335.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.381830310821533,
|
|
"epoch": 0.0075803722504230115,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 6.9661,
|
|
"mean_token_accuracy": 0.08472498953342437,
|
|
"num_tokens": 781500.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.446859312057495,
|
|
"epoch": 0.007715736040609137,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 6.8651,
|
|
"mean_token_accuracy": 0.09278287589550019,
|
|
"num_tokens": 794859.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.3814451694488525,
|
|
"epoch": 0.007851099830795263,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 6.8369,
|
|
"mean_token_accuracy": 0.09373516142368317,
|
|
"num_tokens": 808653.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.418990850448608,
|
|
"epoch": 0.007986463620981387,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.000147,
|
|
"loss": 6.8866,
|
|
"mean_token_accuracy": 0.08324785232543945,
|
|
"num_tokens": 824876.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.358400630950928,
|
|
"epoch": 0.008121827411167513,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 6.8195,
|
|
"mean_token_accuracy": 0.09909781068563461,
|
|
"num_tokens": 838065.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.24797306060791,
|
|
"epoch": 0.008257191201353637,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.000152,
|
|
"loss": 6.7376,
|
|
"mean_token_accuracy": 0.09417691752314568,
|
|
"num_tokens": 850591.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.293615865707397,
|
|
"epoch": 0.008392554991539763,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 6.8553,
|
|
"mean_token_accuracy": 0.09107527285814285,
|
|
"num_tokens": 865254.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.388335704803467,
|
|
"epoch": 0.008527918781725888,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.000157,
|
|
"loss": 6.8402,
|
|
"mean_token_accuracy": 0.08597931787371635,
|
|
"num_tokens": 879791.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.291468143463135,
|
|
"epoch": 0.008663282571912014,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 6.8454,
|
|
"mean_token_accuracy": 0.09367485567927361,
|
|
"num_tokens": 891708.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.272171115875244,
|
|
"epoch": 0.008798646362098138,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.7682,
|
|
"mean_token_accuracy": 0.09533382132649422,
|
|
"num_tokens": 905454.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.248970413208008,
|
|
"epoch": 0.008934010152284264,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.7542,
|
|
"mean_token_accuracy": 0.09669646397233009,
|
|
"num_tokens": 919080.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.180841302871704,
|
|
"epoch": 0.009069373942470389,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 6.6389,
|
|
"mean_token_accuracy": 0.09361326694488525,
|
|
"num_tokens": 932145.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.175373554229736,
|
|
"epoch": 0.009204737732656515,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.7554,
|
|
"mean_token_accuracy": 0.09223678484559059,
|
|
"num_tokens": 946502.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.258382320404053,
|
|
"epoch": 0.009340101522842639,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.7061,
|
|
"mean_token_accuracy": 0.09237500727176666,
|
|
"num_tokens": 959414.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.242776870727539,
|
|
"epoch": 0.009475465313028765,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.7714,
|
|
"mean_token_accuracy": 0.09881808012723922,
|
|
"num_tokens": 972868.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.061698722839355,
|
|
"epoch": 0.00961082910321489,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.6902,
|
|
"mean_token_accuracy": 0.09196390882134438,
|
|
"num_tokens": 987469.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.158710718154907,
|
|
"epoch": 0.009746192893401015,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.693,
|
|
"mean_token_accuracy": 0.09573714062571526,
|
|
"num_tokens": 1001797.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.105884790420532,
|
|
"epoch": 0.00988155668358714,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.6582,
|
|
"mean_token_accuracy": 0.09512505084276199,
|
|
"num_tokens": 1016034.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.184845685958862,
|
|
"epoch": 0.010016920473773266,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.7164,
|
|
"mean_token_accuracy": 0.09947735443711281,
|
|
"num_tokens": 1030566.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.0832836627960205,
|
|
"epoch": 0.01015228426395939,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.763,
|
|
"mean_token_accuracy": 0.09124055355787278,
|
|
"num_tokens": 1044735.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.09069504737854,
|
|
"epoch": 0.010287648054145516,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.7385,
|
|
"mean_token_accuracy": 0.0939220942556858,
|
|
"num_tokens": 1059567.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.149837684631348,
|
|
"epoch": 0.01042301184433164,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.6309,
|
|
"mean_token_accuracy": 0.1013464093208313,
|
|
"num_tokens": 1074612.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.109938526153565,
|
|
"epoch": 0.010558375634517767,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.6941,
|
|
"mean_token_accuracy": 0.09540026187896729,
|
|
"num_tokens": 1087092.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 6.9963634490966795,
|
|
"epoch": 0.010693739424703893,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.6652,
|
|
"mean_token_accuracy": 0.09614972546696662,
|
|
"num_tokens": 1101966.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.059049797058106,
|
|
"epoch": 0.010829103214890017,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.6127,
|
|
"mean_token_accuracy": 0.10141809433698654,
|
|
"num_tokens": 1116287.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.021635675430298,
|
|
"epoch": 0.010964467005076143,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.5773,
|
|
"mean_token_accuracy": 0.10086113512516022,
|
|
"num_tokens": 1132775.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 6.978186845779419,
|
|
"epoch": 0.011099830795262267,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.6211,
|
|
"mean_token_accuracy": 0.09556624442338943,
|
|
"num_tokens": 1149372.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.039114618301392,
|
|
"epoch": 0.011235194585448393,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.7039,
|
|
"mean_token_accuracy": 0.09624231159687043,
|
|
"num_tokens": 1163772.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 6.965072917938232,
|
|
"epoch": 0.011370558375634518,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.5853,
|
|
"mean_token_accuracy": 0.09794198945164681,
|
|
"num_tokens": 1176056.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.030379724502564,
|
|
"epoch": 0.011505922165820644,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.5365,
|
|
"mean_token_accuracy": 0.10276012122631073,
|
|
"num_tokens": 1189759.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.0084771633148195,
|
|
"epoch": 0.011641285956006768,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.591,
|
|
"mean_token_accuracy": 0.10040640980005264,
|
|
"num_tokens": 1203142.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 6.9573570728302006,
|
|
"epoch": 0.011776649746192894,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.5819,
|
|
"mean_token_accuracy": 0.10083244740962982,
|
|
"num_tokens": 1217924.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.0246354103088375,
|
|
"epoch": 0.011912013536379018,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.6666,
|
|
"mean_token_accuracy": 0.10064776018261909,
|
|
"num_tokens": 1231713.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 6.966695070266724,
|
|
"epoch": 0.012047377326565144,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.56,
|
|
"mean_token_accuracy": 0.0989975780248642,
|
|
"num_tokens": 1245538.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 6.837519836425781,
|
|
"epoch": 0.012182741116751269,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.574,
|
|
"mean_token_accuracy": 0.10176394879817963,
|
|
"num_tokens": 1262574.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 6.953103399276733,
|
|
"epoch": 0.012318104906937395,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.4812,
|
|
"mean_token_accuracy": 0.10641162097454071,
|
|
"num_tokens": 1276727.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 6.89174976348877,
|
|
"epoch": 0.01245346869712352,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.4861,
|
|
"mean_token_accuracy": 0.10931641608476639,
|
|
"num_tokens": 1291029.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 6.775276374816895,
|
|
"epoch": 0.012588832487309645,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.4553,
|
|
"mean_token_accuracy": 0.1023554228246212,
|
|
"num_tokens": 1305720.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 6.879015350341797,
|
|
"epoch": 0.01272419627749577,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.4732,
|
|
"mean_token_accuracy": 0.10401393547654152,
|
|
"num_tokens": 1321681.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 6.755443096160889,
|
|
"epoch": 0.012859560067681896,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.4114,
|
|
"mean_token_accuracy": 0.11262697726488113,
|
|
"num_tokens": 1334960.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 6.846341848373413,
|
|
"epoch": 0.01299492385786802,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.58,
|
|
"mean_token_accuracy": 0.10540890470147132,
|
|
"num_tokens": 1349479.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 6.880573081970215,
|
|
"epoch": 0.013130287648054146,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.4669,
|
|
"mean_token_accuracy": 0.10141590088605881,
|
|
"num_tokens": 1363240.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 6.843642616271973,
|
|
"epoch": 0.01326565143824027,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.5119,
|
|
"mean_token_accuracy": 0.09850463718175888,
|
|
"num_tokens": 1375568.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 6.877911949157715,
|
|
"epoch": 0.013401015228426396,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.5538,
|
|
"mean_token_accuracy": 0.10284689888358116,
|
|
"num_tokens": 1389099.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 6.875770854949951,
|
|
"epoch": 0.01353637901861252,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.5636,
|
|
"mean_token_accuracy": 0.09843793138861656,
|
|
"num_tokens": 1402107.0,
|
|
"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": 2465176006656000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|