1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.033488496701383076,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.742561435699463,
|
|
"epoch": 0.00033488496701383075,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7498,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 9336.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.742577743530273,
|
|
"epoch": 0.0006697699340276615,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7384,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 18837.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.742557716369628,
|
|
"epoch": 0.0010046549010414923,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.7272,
|
|
"mean_token_accuracy": 9.718172950670123e-05,
|
|
"num_tokens": 28150.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.742551231384278,
|
|
"epoch": 0.001339539868055323,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.6837,
|
|
"mean_token_accuracy": 0.0009352716442663223,
|
|
"num_tokens": 37469.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.742449760437012,
|
|
"epoch": 0.0016744248350691537,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.6409,
|
|
"mean_token_accuracy": 0.008969881373923272,
|
|
"num_tokens": 47955.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.742036151885987,
|
|
"epoch": 0.0020093098020829846,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.5406,
|
|
"mean_token_accuracy": 0.03436212912201882,
|
|
"num_tokens": 58421.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.740588188171387,
|
|
"epoch": 0.0023441947690968153,
|
|
"grad_norm": 3.0,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.4482,
|
|
"mean_token_accuracy": 0.04009146671742201,
|
|
"num_tokens": 69338.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.735538864135743,
|
|
"epoch": 0.002679079736110646,
|
|
"grad_norm": 2.5625,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.3231,
|
|
"mean_token_accuracy": 0.0391630582511425,
|
|
"num_tokens": 78964.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.725551223754882,
|
|
"epoch": 0.0030139647031244767,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.2374,
|
|
"mean_token_accuracy": 0.046432715654373166,
|
|
"num_tokens": 90312.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.711534595489502,
|
|
"epoch": 0.0033488496701383074,
|
|
"grad_norm": 2.078125,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 10.1706,
|
|
"mean_token_accuracy": 0.040825183317065236,
|
|
"num_tokens": 99706.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.701226234436035,
|
|
"epoch": 0.003683734637152138,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 10.1097,
|
|
"mean_token_accuracy": 0.04384128861129284,
|
|
"num_tokens": 109737.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.695288467407227,
|
|
"epoch": 0.004018619604165969,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 10.0916,
|
|
"mean_token_accuracy": 0.04214213602244854,
|
|
"num_tokens": 119471.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.689713382720948,
|
|
"epoch": 0.0043535045711798,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 10.0375,
|
|
"mean_token_accuracy": 0.041708166152238844,
|
|
"num_tokens": 130844.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.684234142303467,
|
|
"epoch": 0.0046883895381936306,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.9974,
|
|
"mean_token_accuracy": 0.04208777397871018,
|
|
"num_tokens": 141266.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.676689434051514,
|
|
"epoch": 0.005023274505207461,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.898,
|
|
"mean_token_accuracy": 0.04250998180359602,
|
|
"num_tokens": 152222.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.664115810394287,
|
|
"epoch": 0.005358159472221292,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.8609,
|
|
"mean_token_accuracy": 0.04332525990903378,
|
|
"num_tokens": 162931.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.650795078277588,
|
|
"epoch": 0.005693044439235123,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.8387,
|
|
"mean_token_accuracy": 0.044832929782569406,
|
|
"num_tokens": 172587.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.638714599609376,
|
|
"epoch": 0.006027929406248953,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.768,
|
|
"mean_token_accuracy": 0.04449521526694298,
|
|
"num_tokens": 181877.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.623728466033935,
|
|
"epoch": 0.006362814373262784,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.6987,
|
|
"mean_token_accuracy": 0.047365859150886536,
|
|
"num_tokens": 192196.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.59886417388916,
|
|
"epoch": 0.006697699340276615,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.6205,
|
|
"mean_token_accuracy": 0.04793250001966953,
|
|
"num_tokens": 201009.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.578139305114746,
|
|
"epoch": 0.007032584307290445,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.5589,
|
|
"mean_token_accuracy": 0.05401066094636917,
|
|
"num_tokens": 209066.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.551801109313965,
|
|
"epoch": 0.007367469274304276,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.4901,
|
|
"mean_token_accuracy": 0.05924240723252296,
|
|
"num_tokens": 218721.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.502099704742431,
|
|
"epoch": 0.007702354241318108,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 9.4194,
|
|
"mean_token_accuracy": 0.05640989877283573,
|
|
"num_tokens": 228511.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.450156116485596,
|
|
"epoch": 0.008037239208331938,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 9.3421,
|
|
"mean_token_accuracy": 0.060460330545902254,
|
|
"num_tokens": 237097.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.362788772583007,
|
|
"epoch": 0.008372124175345769,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 9.2305,
|
|
"mean_token_accuracy": 0.06126848310232162,
|
|
"num_tokens": 246013.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.28974199295044,
|
|
"epoch": 0.0087070091423596,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 9.2132,
|
|
"mean_token_accuracy": 0.0567143976688385,
|
|
"num_tokens": 255741.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.277145481109619,
|
|
"epoch": 0.00904189410937343,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 9.0746,
|
|
"mean_token_accuracy": 0.060845992714166644,
|
|
"num_tokens": 265877.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.188594436645507,
|
|
"epoch": 0.009376779076387261,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 9.0594,
|
|
"mean_token_accuracy": 0.05822984836995602,
|
|
"num_tokens": 275631.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.113962650299072,
|
|
"epoch": 0.009711664043401092,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.9686,
|
|
"mean_token_accuracy": 0.06056818813085556,
|
|
"num_tokens": 286584.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.048769855499268,
|
|
"epoch": 0.010046549010414923,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.9176,
|
|
"mean_token_accuracy": 0.05582499206066131,
|
|
"num_tokens": 296903.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.948976993560791,
|
|
"epoch": 0.010381433977428753,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.8091,
|
|
"mean_token_accuracy": 0.058087332919239995,
|
|
"num_tokens": 306940.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.834512996673585,
|
|
"epoch": 0.010716318944442584,
|
|
"grad_norm": 0.78125,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 8.821,
|
|
"mean_token_accuracy": 0.05299450382590294,
|
|
"num_tokens": 317344.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.730674839019775,
|
|
"epoch": 0.011051203911456415,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 8.6657,
|
|
"mean_token_accuracy": 0.06257540844380856,
|
|
"num_tokens": 326076.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.602271366119385,
|
|
"epoch": 0.011386088878470245,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 8.6223,
|
|
"mean_token_accuracy": 0.05909014530479908,
|
|
"num_tokens": 334645.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.467283821105957,
|
|
"epoch": 0.011720973845484076,
|
|
"grad_norm": 0.87890625,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 8.6422,
|
|
"mean_token_accuracy": 0.06433657817542553,
|
|
"num_tokens": 343784.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 9.435225105285644,
|
|
"epoch": 0.012055858812497907,
|
|
"grad_norm": 0.83203125,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 8.5783,
|
|
"mean_token_accuracy": 0.05760964490473271,
|
|
"num_tokens": 353230.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 9.283469009399415,
|
|
"epoch": 0.012390743779511737,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 8.5425,
|
|
"mean_token_accuracy": 0.06422759890556336,
|
|
"num_tokens": 362406.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 9.118220233917237,
|
|
"epoch": 0.012725628746525568,
|
|
"grad_norm": 0.8046875,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 8.4932,
|
|
"mean_token_accuracy": 0.06339454613626003,
|
|
"num_tokens": 372418.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 9.104973220825196,
|
|
"epoch": 0.013060513713539399,
|
|
"grad_norm": 0.8828125,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 8.484,
|
|
"mean_token_accuracy": 0.06246692761778831,
|
|
"num_tokens": 381089.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 9.07259292602539,
|
|
"epoch": 0.01339539868055323,
|
|
"grad_norm": 0.7421875,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 8.5434,
|
|
"mean_token_accuracy": 0.05604027174413204,
|
|
"num_tokens": 391801.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 9.01559944152832,
|
|
"epoch": 0.01373028364756706,
|
|
"grad_norm": 0.80859375,
|
|
"learning_rate": 0.000102,
|
|
"loss": 8.5002,
|
|
"mean_token_accuracy": 0.06068759709596634,
|
|
"num_tokens": 402234.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.936877250671387,
|
|
"epoch": 0.01406516861458089,
|
|
"grad_norm": 0.8125,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 8.479,
|
|
"mean_token_accuracy": 0.06238379329442978,
|
|
"num_tokens": 411665.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.94469394683838,
|
|
"epoch": 0.014400053581594721,
|
|
"grad_norm": 0.76171875,
|
|
"learning_rate": 0.000107,
|
|
"loss": 8.4931,
|
|
"mean_token_accuracy": 0.0650034338235855,
|
|
"num_tokens": 421989.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.900172710418701,
|
|
"epoch": 0.014734938548608552,
|
|
"grad_norm": 0.796875,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 8.3895,
|
|
"mean_token_accuracy": 0.06547467894852162,
|
|
"num_tokens": 431258.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.874738311767578,
|
|
"epoch": 0.015069823515622383,
|
|
"grad_norm": 0.7421875,
|
|
"learning_rate": 0.000112,
|
|
"loss": 8.4852,
|
|
"mean_token_accuracy": 0.06038682572543621,
|
|
"num_tokens": 441920.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.890042400360107,
|
|
"epoch": 0.015404708482636215,
|
|
"grad_norm": 0.7890625,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 8.3921,
|
|
"mean_token_accuracy": 0.06986854746937751,
|
|
"num_tokens": 451460.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 8.85048770904541,
|
|
"epoch": 0.015739593449650046,
|
|
"grad_norm": 0.828125,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 8.3998,
|
|
"mean_token_accuracy": 0.06450125053524972,
|
|
"num_tokens": 461612.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.77711887359619,
|
|
"epoch": 0.016074478416663877,
|
|
"grad_norm": 0.85546875,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 8.4376,
|
|
"mean_token_accuracy": 0.06631115414202213,
|
|
"num_tokens": 472706.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 8.85504970550537,
|
|
"epoch": 0.016409363383677707,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.000122,
|
|
"loss": 8.4614,
|
|
"mean_token_accuracy": 0.061580969393253325,
|
|
"num_tokens": 482709.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.814317607879639,
|
|
"epoch": 0.016744248350691538,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 8.4475,
|
|
"mean_token_accuracy": 0.06584244705736637,
|
|
"num_tokens": 492566.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.831695365905762,
|
|
"epoch": 0.01707913331770537,
|
|
"grad_norm": 0.88671875,
|
|
"learning_rate": 0.000127,
|
|
"loss": 8.4019,
|
|
"mean_token_accuracy": 0.06290778964757919,
|
|
"num_tokens": 502912.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 8.831443119049073,
|
|
"epoch": 0.0174140182847192,
|
|
"grad_norm": 0.8515625,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 8.4192,
|
|
"mean_token_accuracy": 0.06170829311013222,
|
|
"num_tokens": 513000.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 8.78964204788208,
|
|
"epoch": 0.01774890325173303,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.000132,
|
|
"loss": 8.3577,
|
|
"mean_token_accuracy": 0.06793267726898193,
|
|
"num_tokens": 523345.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 8.735497283935548,
|
|
"epoch": 0.01808378821874686,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 8.3634,
|
|
"mean_token_accuracy": 0.07245749980211258,
|
|
"num_tokens": 532121.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 8.812878513336182,
|
|
"epoch": 0.01841867318576069,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 8.4324,
|
|
"mean_token_accuracy": 0.06445505432784557,
|
|
"num_tokens": 541895.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 8.757740497589111,
|
|
"epoch": 0.018753558152774522,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 8.3649,
|
|
"mean_token_accuracy": 0.06517385765910148,
|
|
"num_tokens": 552144.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 8.80039710998535,
|
|
"epoch": 0.019088443119788353,
|
|
"grad_norm": 0.81640625,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 8.3675,
|
|
"mean_token_accuracy": 0.06585304848849774,
|
|
"num_tokens": 562336.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 8.644570541381835,
|
|
"epoch": 0.019423328086802184,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 8.3217,
|
|
"mean_token_accuracy": 0.07459635213017464,
|
|
"num_tokens": 572020.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 8.7322265625,
|
|
"epoch": 0.019758213053816014,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.000147,
|
|
"loss": 8.3229,
|
|
"mean_token_accuracy": 0.07014562897384166,
|
|
"num_tokens": 581418.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 8.696172523498536,
|
|
"epoch": 0.020093098020829845,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 8.304,
|
|
"mean_token_accuracy": 0.07059184238314628,
|
|
"num_tokens": 590716.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 8.63413028717041,
|
|
"epoch": 0.020427982987843676,
|
|
"grad_norm": 0.8515625,
|
|
"learning_rate": 0.000152,
|
|
"loss": 8.2026,
|
|
"mean_token_accuracy": 0.07757920697331429,
|
|
"num_tokens": 599337.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 8.697781085968018,
|
|
"epoch": 0.020762867954857506,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 8.3118,
|
|
"mean_token_accuracy": 0.06973933763802051,
|
|
"num_tokens": 609600.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 8.631188106536865,
|
|
"epoch": 0.021097752921871337,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000157,
|
|
"loss": 8.2052,
|
|
"mean_token_accuracy": 0.0789581760764122,
|
|
"num_tokens": 619121.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 8.685782337188721,
|
|
"epoch": 0.021432637888885168,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 8.2304,
|
|
"mean_token_accuracy": 0.07488703206181527,
|
|
"num_tokens": 628741.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 8.586778736114502,
|
|
"epoch": 0.021767522855899,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.000162,
|
|
"loss": 8.3125,
|
|
"mean_token_accuracy": 0.07511252909898758,
|
|
"num_tokens": 637422.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 8.691543483734131,
|
|
"epoch": 0.02210240782291283,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 8.3278,
|
|
"mean_token_accuracy": 0.06828603111207485,
|
|
"num_tokens": 646541.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 8.629717159271241,
|
|
"epoch": 0.02243729278992666,
|
|
"grad_norm": 0.8203125,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 8.2181,
|
|
"mean_token_accuracy": 0.07183332405984402,
|
|
"num_tokens": 656771.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 8.613917541503906,
|
|
"epoch": 0.02277217775694049,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 8.2728,
|
|
"mean_token_accuracy": 0.08168347477912903,
|
|
"num_tokens": 666273.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 8.618037605285645,
|
|
"epoch": 0.02310706272395432,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 8.2002,
|
|
"mean_token_accuracy": 0.07752258330583572,
|
|
"num_tokens": 676473.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 8.643811130523682,
|
|
"epoch": 0.023441947690968152,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 8.273,
|
|
"mean_token_accuracy": 0.07563822567462922,
|
|
"num_tokens": 686180.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 8.547807312011718,
|
|
"epoch": 0.023776832657981983,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.000177,
|
|
"loss": 8.2201,
|
|
"mean_token_accuracy": 0.07755913883447647,
|
|
"num_tokens": 695372.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 8.731934547424316,
|
|
"epoch": 0.024111717624995813,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 8.3056,
|
|
"mean_token_accuracy": 0.07374768368899823,
|
|
"num_tokens": 706182.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 8.546996784210204,
|
|
"epoch": 0.024446602592009644,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.000182,
|
|
"loss": 8.1937,
|
|
"mean_token_accuracy": 0.08396602459251881,
|
|
"num_tokens": 715601.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 8.672658920288086,
|
|
"epoch": 0.024781487559023475,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 8.281,
|
|
"mean_token_accuracy": 0.07340774051845074,
|
|
"num_tokens": 724919.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 8.621439933776855,
|
|
"epoch": 0.025116372526037305,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000187,
|
|
"loss": 8.1873,
|
|
"mean_token_accuracy": 0.08298620432615281,
|
|
"num_tokens": 733214.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 8.571719074249268,
|
|
"epoch": 0.025451257493051136,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 8.2474,
|
|
"mean_token_accuracy": 0.07782503142952919,
|
|
"num_tokens": 743230.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 8.560567378997803,
|
|
"epoch": 0.025786142460064967,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 8.1869,
|
|
"mean_token_accuracy": 0.08309442102909088,
|
|
"num_tokens": 753175.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 8.578225898742676,
|
|
"epoch": 0.026121027427078797,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 8.1224,
|
|
"mean_token_accuracy": 0.0769454549998045,
|
|
"num_tokens": 763168.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 8.538741397857667,
|
|
"epoch": 0.026455912394092628,
|
|
"grad_norm": 0.8046875,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 8.2106,
|
|
"mean_token_accuracy": 0.0778761513531208,
|
|
"num_tokens": 773477.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 8.53778133392334,
|
|
"epoch": 0.02679079736110646,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 8.2055,
|
|
"mean_token_accuracy": 0.07703198343515397,
|
|
"num_tokens": 782351.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 8.543316268920899,
|
|
"epoch": 0.02712568232812029,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.000202,
|
|
"loss": 8.1518,
|
|
"mean_token_accuracy": 0.08180877156555652,
|
|
"num_tokens": 792179.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 8.528862571716308,
|
|
"epoch": 0.02746056729513412,
|
|
"grad_norm": 0.86328125,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 8.1353,
|
|
"mean_token_accuracy": 0.07953180633485317,
|
|
"num_tokens": 802243.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 8.574699401855469,
|
|
"epoch": 0.02779545226214795,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000207,
|
|
"loss": 8.1209,
|
|
"mean_token_accuracy": 0.08045852780342103,
|
|
"num_tokens": 812069.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 8.437831401824951,
|
|
"epoch": 0.02813033722916178,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 8.1081,
|
|
"mean_token_accuracy": 0.08233479075133801,
|
|
"num_tokens": 821608.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 8.531076431274414,
|
|
"epoch": 0.028465222196175612,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.000212,
|
|
"loss": 8.2246,
|
|
"mean_token_accuracy": 0.07980309575796127,
|
|
"num_tokens": 831677.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 8.510465621948242,
|
|
"epoch": 0.028800107163189443,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 8.0808,
|
|
"mean_token_accuracy": 0.08042851686477662,
|
|
"num_tokens": 840087.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 8.538240432739258,
|
|
"epoch": 0.029134992130203274,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 8.0842,
|
|
"mean_token_accuracy": 0.08349445015192032,
|
|
"num_tokens": 850516.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 8.433059692382812,
|
|
"epoch": 0.029469877097217104,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 8.1666,
|
|
"mean_token_accuracy": 0.07715897038578987,
|
|
"num_tokens": 860921.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 8.488123321533203,
|
|
"epoch": 0.029804762064230935,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.000222,
|
|
"loss": 8.0755,
|
|
"mean_token_accuracy": 0.07828016802668572,
|
|
"num_tokens": 871450.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 8.392322254180907,
|
|
"epoch": 0.030139647031244766,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 8.0472,
|
|
"mean_token_accuracy": 0.08468902185559272,
|
|
"num_tokens": 880241.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 8.428736782073974,
|
|
"epoch": 0.030474531998258596,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 8.0976,
|
|
"mean_token_accuracy": 0.08420571461319923,
|
|
"num_tokens": 890156.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 8.475104236602784,
|
|
"epoch": 0.03080941696527243,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 8.1217,
|
|
"mean_token_accuracy": 0.08020563796162605,
|
|
"num_tokens": 899718.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 8.438492584228516,
|
|
"epoch": 0.03114430193228626,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 8.0464,
|
|
"mean_token_accuracy": 0.08637708500027656,
|
|
"num_tokens": 908461.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 8.375166416168213,
|
|
"epoch": 0.03147918689930009,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 7.9991,
|
|
"mean_token_accuracy": 0.0926103100180626,
|
|
"num_tokens": 917418.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 8.406765747070313,
|
|
"epoch": 0.03181407186631392,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.000237,
|
|
"loss": 8.079,
|
|
"mean_token_accuracy": 0.08318312540650367,
|
|
"num_tokens": 927495.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 8.399221515655517,
|
|
"epoch": 0.03214895683332775,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 8.0856,
|
|
"mean_token_accuracy": 0.08131228312849999,
|
|
"num_tokens": 937510.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 8.47674627304077,
|
|
"epoch": 0.032483841800341584,
|
|
"grad_norm": 0.875,
|
|
"learning_rate": 0.000242,
|
|
"loss": 8.0468,
|
|
"mean_token_accuracy": 0.08284490816295147,
|
|
"num_tokens": 948171.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 8.403480339050294,
|
|
"epoch": 0.032818726767355415,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 8.0069,
|
|
"mean_token_accuracy": 0.08470370024442672,
|
|
"num_tokens": 956506.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 8.396288394927979,
|
|
"epoch": 0.033153611734369245,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000247,
|
|
"loss": 8.0398,
|
|
"mean_token_accuracy": 0.08274327740073203,
|
|
"num_tokens": 966063.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 8.39925765991211,
|
|
"epoch": 0.033488496701383076,
|
|
"grad_norm": 0.84375,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 8.0258,
|
|
"mean_token_accuracy": 0.08637790307402611,
|
|
"num_tokens": 976793.0,
|
|
"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": 226186232463360.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|