1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.027977506085107573,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.742612838745117,
|
|
"epoch": 0.0002797750608510757,
|
|
"grad_norm": 5.8125,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7438,
|
|
"mean_token_accuracy": 0.00013192612677812577,
|
|
"num_tokens": 8714.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.74260196685791,
|
|
"epoch": 0.0005595501217021514,
|
|
"grad_norm": 5.0625,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7505,
|
|
"mean_token_accuracy": 6.393861840479076e-05,
|
|
"num_tokens": 18630.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.742606258392334,
|
|
"epoch": 0.0008393251825532272,
|
|
"grad_norm": 5.25,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.7114,
|
|
"mean_token_accuracy": 0.00019184346892870964,
|
|
"num_tokens": 28134.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.74264087677002,
|
|
"epoch": 0.0011191002434043028,
|
|
"grad_norm": 5.03125,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.6643,
|
|
"mean_token_accuracy": 0.00022531848517246545,
|
|
"num_tokens": 37326.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.742629432678223,
|
|
"epoch": 0.0013988753042553786,
|
|
"grad_norm": 4.8125,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.5505,
|
|
"mean_token_accuracy": 0.006841035454999656,
|
|
"num_tokens": 45813.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.74232587814331,
|
|
"epoch": 0.0016786503651064545,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.4343,
|
|
"mean_token_accuracy": 0.029637257009744643,
|
|
"num_tokens": 54746.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.74130563735962,
|
|
"epoch": 0.0019584254259575303,
|
|
"grad_norm": 3.21875,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.2946,
|
|
"mean_token_accuracy": 0.044694996997714044,
|
|
"num_tokens": 64386.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.73903694152832,
|
|
"epoch": 0.0022382004868086057,
|
|
"grad_norm": 2.71875,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.1927,
|
|
"mean_token_accuracy": 0.0362966725602746,
|
|
"num_tokens": 73648.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.735822677612305,
|
|
"epoch": 0.0025179755476596815,
|
|
"grad_norm": 2.3125,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.0485,
|
|
"mean_token_accuracy": 0.03808995056897402,
|
|
"num_tokens": 82108.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.732699203491212,
|
|
"epoch": 0.0027977506085107573,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.9639,
|
|
"mean_token_accuracy": 0.04292008355259895,
|
|
"num_tokens": 92109.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.730623531341553,
|
|
"epoch": 0.003077525669361833,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.8893,
|
|
"mean_token_accuracy": 0.039536670222878455,
|
|
"num_tokens": 101365.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.72904348373413,
|
|
"epoch": 0.003357300730212909,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.8475,
|
|
"mean_token_accuracy": 0.04103112425655127,
|
|
"num_tokens": 110432.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.726887702941895,
|
|
"epoch": 0.0036370757910639847,
|
|
"grad_norm": 2.0,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.8013,
|
|
"mean_token_accuracy": 0.04199505355209112,
|
|
"num_tokens": 120191.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.72308177947998,
|
|
"epoch": 0.0039168508519150606,
|
|
"grad_norm": 1.9921875,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.697,
|
|
"mean_token_accuracy": 0.04964672140777111,
|
|
"num_tokens": 129580.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.715501308441162,
|
|
"epoch": 0.004196625912766136,
|
|
"grad_norm": 1.8203125,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.6931,
|
|
"mean_token_accuracy": 0.04204687792807817,
|
|
"num_tokens": 139083.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.71017141342163,
|
|
"epoch": 0.004476400973617211,
|
|
"grad_norm": 1.9375,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.547,
|
|
"mean_token_accuracy": 0.044005101546645164,
|
|
"num_tokens": 148183.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.70678539276123,
|
|
"epoch": 0.004756176034468288,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.5113,
|
|
"mean_token_accuracy": 0.044296158105134965,
|
|
"num_tokens": 157015.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.700459957122803,
|
|
"epoch": 0.005035951095319363,
|
|
"grad_norm": 1.875,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.4495,
|
|
"mean_token_accuracy": 0.04853829797357321,
|
|
"num_tokens": 166314.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.690591049194335,
|
|
"epoch": 0.005315726156170439,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.3275,
|
|
"mean_token_accuracy": 0.056649215146899225,
|
|
"num_tokens": 175975.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.673609352111816,
|
|
"epoch": 0.005595501217021515,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.2483,
|
|
"mean_token_accuracy": 0.05905318446457386,
|
|
"num_tokens": 185976.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.651649284362794,
|
|
"epoch": 0.005875276277872591,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.1583,
|
|
"mean_token_accuracy": 0.053412462770938876,
|
|
"num_tokens": 195112.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.622291564941406,
|
|
"epoch": 0.006155051338723666,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.0868,
|
|
"mean_token_accuracy": 0.05187728106975555,
|
|
"num_tokens": 205951.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.582518100738525,
|
|
"epoch": 0.006434826399574742,
|
|
"grad_norm": 1.7890625,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.9334,
|
|
"mean_token_accuracy": 0.05757487155497074,
|
|
"num_tokens": 214744.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.516531848907471,
|
|
"epoch": 0.006714601460425818,
|
|
"grad_norm": 1.6953125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.785,
|
|
"mean_token_accuracy": 0.054786504805088045,
|
|
"num_tokens": 223279.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.463547229766846,
|
|
"epoch": 0.006994376521276893,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.7489,
|
|
"mean_token_accuracy": 0.05046164020895958,
|
|
"num_tokens": 232677.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.413999271392822,
|
|
"epoch": 0.0072741515821279695,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.5832,
|
|
"mean_token_accuracy": 0.053359159454703334,
|
|
"num_tokens": 242890.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.342116355895996,
|
|
"epoch": 0.007553926642979045,
|
|
"grad_norm": 1.703125,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.4072,
|
|
"mean_token_accuracy": 0.05878838151693344,
|
|
"num_tokens": 252490.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.246252346038819,
|
|
"epoch": 0.007833701703830121,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.3198,
|
|
"mean_token_accuracy": 0.05760062858462334,
|
|
"num_tokens": 261280.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.14638671875,
|
|
"epoch": 0.008113476764681197,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.2392,
|
|
"mean_token_accuracy": 0.0589225921779871,
|
|
"num_tokens": 270000.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.029090118408202,
|
|
"epoch": 0.008393251825532272,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.0511,
|
|
"mean_token_accuracy": 0.06310832649469375,
|
|
"num_tokens": 279709.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.880472183227539,
|
|
"epoch": 0.008673026886383347,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.0369,
|
|
"mean_token_accuracy": 0.05967277251183987,
|
|
"num_tokens": 288343.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.730181789398193,
|
|
"epoch": 0.008952801947234423,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.8578,
|
|
"mean_token_accuracy": 0.06264004111289978,
|
|
"num_tokens": 297387.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.494822692871093,
|
|
"epoch": 0.0092325770080855,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.7403,
|
|
"mean_token_accuracy": 0.0650530807673931,
|
|
"num_tokens": 307175.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.333676815032959,
|
|
"epoch": 0.009512352068936575,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.6329,
|
|
"mean_token_accuracy": 0.06255011968314647,
|
|
"num_tokens": 316033.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.095909214019775,
|
|
"epoch": 0.00979212712978765,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.6513,
|
|
"mean_token_accuracy": 0.06538875326514244,
|
|
"num_tokens": 325740.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.91105089187622,
|
|
"epoch": 0.010071902190638726,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.5263,
|
|
"mean_token_accuracy": 0.0705183681100607,
|
|
"num_tokens": 334637.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.67013111114502,
|
|
"epoch": 0.010351677251489803,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.5331,
|
|
"mean_token_accuracy": 0.06915191635489464,
|
|
"num_tokens": 344020.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.54305191040039,
|
|
"epoch": 0.010631452312340878,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.3922,
|
|
"mean_token_accuracy": 0.07512268126010894,
|
|
"num_tokens": 352924.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.426819992065429,
|
|
"epoch": 0.010911227373191954,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.3884,
|
|
"mean_token_accuracy": 0.07741571851074695,
|
|
"num_tokens": 361755.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.332173538208007,
|
|
"epoch": 0.01119100243404303,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.4178,
|
|
"mean_token_accuracy": 0.07555796578526497,
|
|
"num_tokens": 370605.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.264447212219238,
|
|
"epoch": 0.011470777494894105,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.2953,
|
|
"mean_token_accuracy": 0.08062206991016865,
|
|
"num_tokens": 379723.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.139957714080811,
|
|
"epoch": 0.011750552555745182,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.2292,
|
|
"mean_token_accuracy": 0.07732874378561974,
|
|
"num_tokens": 389502.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.078060913085938,
|
|
"epoch": 0.012030327616596257,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.2085,
|
|
"mean_token_accuracy": 0.07959119230508804,
|
|
"num_tokens": 399116.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.096036434173584,
|
|
"epoch": 0.012310102677447332,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.249,
|
|
"mean_token_accuracy": 0.07255740389227867,
|
|
"num_tokens": 408239.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.055156040191651,
|
|
"epoch": 0.012589877738298408,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.2475,
|
|
"mean_token_accuracy": 0.08420794308185578,
|
|
"num_tokens": 418801.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.940054512023925,
|
|
"epoch": 0.012869652799149483,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.2634,
|
|
"mean_token_accuracy": 0.08109545595943927,
|
|
"num_tokens": 427584.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.987145185470581,
|
|
"epoch": 0.01314942786000056,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.2117,
|
|
"mean_token_accuracy": 0.08332281783223153,
|
|
"num_tokens": 437345.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.946065521240234,
|
|
"epoch": 0.013429202920851636,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.2381,
|
|
"mean_token_accuracy": 0.08383557200431824,
|
|
"num_tokens": 446550.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.836467123031616,
|
|
"epoch": 0.013708977981702711,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.2057,
|
|
"mean_token_accuracy": 0.078835878521204,
|
|
"num_tokens": 456394.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.929483318328858,
|
|
"epoch": 0.013988753042553786,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.0898,
|
|
"mean_token_accuracy": 0.08839634135365486,
|
|
"num_tokens": 465467.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.76855092048645,
|
|
"epoch": 0.014268528103404862,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.1608,
|
|
"mean_token_accuracy": 0.08721910566091537,
|
|
"num_tokens": 474838.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.877699708938598,
|
|
"epoch": 0.014548303164255939,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.0455,
|
|
"mean_token_accuracy": 0.09059784710407257,
|
|
"num_tokens": 483414.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.812541151046753,
|
|
"epoch": 0.014828078225107014,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.0716,
|
|
"mean_token_accuracy": 0.09124693721532821,
|
|
"num_tokens": 491802.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.714865016937256,
|
|
"epoch": 0.01510785328595809,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.0616,
|
|
"mean_token_accuracy": 0.08571952506899834,
|
|
"num_tokens": 502007.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.775686359405517,
|
|
"epoch": 0.015387628346809165,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.0939,
|
|
"mean_token_accuracy": 0.08662015125155449,
|
|
"num_tokens": 510806.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.638946437835694,
|
|
"epoch": 0.015667403407660242,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.0374,
|
|
"mean_token_accuracy": 0.08706805557012558,
|
|
"num_tokens": 519526.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.687864446640015,
|
|
"epoch": 0.015947178468511316,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 6.9481,
|
|
"mean_token_accuracy": 0.09310803934931755,
|
|
"num_tokens": 529036.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.70524320602417,
|
|
"epoch": 0.016226953529362393,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.023,
|
|
"mean_token_accuracy": 0.08589281886816025,
|
|
"num_tokens": 537758.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.609296417236328,
|
|
"epoch": 0.016506728590213467,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000147,
|
|
"loss": 6.9372,
|
|
"mean_token_accuracy": 0.09534673020243645,
|
|
"num_tokens": 546401.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.591969633102417,
|
|
"epoch": 0.016786503651064544,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 6.9473,
|
|
"mean_token_accuracy": 0.08842713311314583,
|
|
"num_tokens": 555839.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.720137023925782,
|
|
"epoch": 0.01706627871191562,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.0974,
|
|
"mean_token_accuracy": 0.09283825233578683,
|
|
"num_tokens": 565150.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.533408880233765,
|
|
"epoch": 0.017346053772766695,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 6.9415,
|
|
"mean_token_accuracy": 0.09149915426969528,
|
|
"num_tokens": 574376.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.585634994506836,
|
|
"epoch": 0.01762582883361777,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.0738,
|
|
"mean_token_accuracy": 0.08485013097524643,
|
|
"num_tokens": 583075.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.657504510879517,
|
|
"epoch": 0.017905603894468845,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.0915,
|
|
"mean_token_accuracy": 0.08915092051029205,
|
|
"num_tokens": 593003.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.495992803573609,
|
|
"epoch": 0.018185378955319922,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.8975,
|
|
"mean_token_accuracy": 0.09913151040673256,
|
|
"num_tokens": 602273.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.5305685043334964,
|
|
"epoch": 0.018465154016171,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.8442,
|
|
"mean_token_accuracy": 0.09180534854531289,
|
|
"num_tokens": 610364.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.406687688827515,
|
|
"epoch": 0.018744929077022073,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 6.8416,
|
|
"mean_token_accuracy": 0.09668334275484085,
|
|
"num_tokens": 618753.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.572203159332275,
|
|
"epoch": 0.01902470413787315,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.9437,
|
|
"mean_token_accuracy": 0.09190804958343506,
|
|
"num_tokens": 627697.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.491861772537232,
|
|
"epoch": 0.019304479198724227,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.9013,
|
|
"mean_token_accuracy": 0.0921478547155857,
|
|
"num_tokens": 636499.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.469914817810059,
|
|
"epoch": 0.0195842542595753,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.916,
|
|
"mean_token_accuracy": 0.08606895133852958,
|
|
"num_tokens": 646426.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.453820514678955,
|
|
"epoch": 0.019864029320426378,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.848,
|
|
"mean_token_accuracy": 0.09033864736557007,
|
|
"num_tokens": 656593.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.423789358139038,
|
|
"epoch": 0.020143804381277452,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.8855,
|
|
"mean_token_accuracy": 0.09489151760935784,
|
|
"num_tokens": 665631.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.381592226028443,
|
|
"epoch": 0.02042357944212853,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.8033,
|
|
"mean_token_accuracy": 0.09809573590755463,
|
|
"num_tokens": 674943.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.3162353992462155,
|
|
"epoch": 0.020703354502979606,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.711,
|
|
"mean_token_accuracy": 0.102563014626503,
|
|
"num_tokens": 683500.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.442006301879883,
|
|
"epoch": 0.02098312956383068,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.9107,
|
|
"mean_token_accuracy": 0.08907283470034599,
|
|
"num_tokens": 692147.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.394319009780884,
|
|
"epoch": 0.021262904624681757,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.8442,
|
|
"mean_token_accuracy": 0.09654413685202598,
|
|
"num_tokens": 701226.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.378800916671753,
|
|
"epoch": 0.02154267968553283,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.8664,
|
|
"mean_token_accuracy": 0.09722913801670074,
|
|
"num_tokens": 710886.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.401981687545776,
|
|
"epoch": 0.021822454746383908,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.8006,
|
|
"mean_token_accuracy": 0.09455461874604225,
|
|
"num_tokens": 719209.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.255378770828247,
|
|
"epoch": 0.022102229807234985,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.6876,
|
|
"mean_token_accuracy": 0.10579567775130272,
|
|
"num_tokens": 728376.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.269871187210083,
|
|
"epoch": 0.02238200486808606,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.7438,
|
|
"mean_token_accuracy": 0.0953303910791874,
|
|
"num_tokens": 737110.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.368067264556885,
|
|
"epoch": 0.022661779928937135,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.7965,
|
|
"mean_token_accuracy": 0.0932080514729023,
|
|
"num_tokens": 746133.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.259396600723266,
|
|
"epoch": 0.02294155498978821,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.7398,
|
|
"mean_token_accuracy": 0.10008203461766244,
|
|
"num_tokens": 755994.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.206631469726562,
|
|
"epoch": 0.023221330050639286,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.7016,
|
|
"mean_token_accuracy": 0.101080971211195,
|
|
"num_tokens": 764630.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.228565454483032,
|
|
"epoch": 0.023501105111490363,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.7274,
|
|
"mean_token_accuracy": 0.10294575095176697,
|
|
"num_tokens": 773577.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.299652767181397,
|
|
"epoch": 0.023780880172341437,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.6283,
|
|
"mean_token_accuracy": 0.10195705965161324,
|
|
"num_tokens": 782405.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.235974836349487,
|
|
"epoch": 0.024060655233192514,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.7845,
|
|
"mean_token_accuracy": 0.10176606699824334,
|
|
"num_tokens": 790940.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.185374164581299,
|
|
"epoch": 0.024340430294043588,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.7722,
|
|
"mean_token_accuracy": 0.09813026785850525,
|
|
"num_tokens": 800023.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.329893302917481,
|
|
"epoch": 0.024620205354894665,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.6582,
|
|
"mean_token_accuracy": 0.1064946822822094,
|
|
"num_tokens": 809146.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.11688723564148,
|
|
"epoch": 0.024899980415745742,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.649,
|
|
"mean_token_accuracy": 0.10224289745092392,
|
|
"num_tokens": 818300.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.139934825897217,
|
|
"epoch": 0.025179755476596816,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.6744,
|
|
"mean_token_accuracy": 0.10481091514229775,
|
|
"num_tokens": 826991.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.184496831893921,
|
|
"epoch": 0.025459530537447893,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.6768,
|
|
"mean_token_accuracy": 0.10479119941592216,
|
|
"num_tokens": 836530.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.223911333084106,
|
|
"epoch": 0.025739305598298966,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.7759,
|
|
"mean_token_accuracy": 0.10380629897117614,
|
|
"num_tokens": 845754.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.119731903076172,
|
|
"epoch": 0.026019080659150044,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.6838,
|
|
"mean_token_accuracy": 0.10599191337823868,
|
|
"num_tokens": 854778.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.322100830078125,
|
|
"epoch": 0.02629885572000112,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.6936,
|
|
"mean_token_accuracy": 0.1073679007589817,
|
|
"num_tokens": 863057.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.157601404190063,
|
|
"epoch": 0.026578630780852194,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.7808,
|
|
"mean_token_accuracy": 0.09569477587938309,
|
|
"num_tokens": 872781.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.127511215209961,
|
|
"epoch": 0.02685840584170327,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.5325,
|
|
"mean_token_accuracy": 0.1027837336063385,
|
|
"num_tokens": 881815.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.074372053146362,
|
|
"epoch": 0.027138180902554345,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.6255,
|
|
"mean_token_accuracy": 0.10499309822916984,
|
|
"num_tokens": 890745.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.085717725753784,
|
|
"epoch": 0.027417955963405422,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.6589,
|
|
"mean_token_accuracy": 0.10791194960474967,
|
|
"num_tokens": 900315.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.077784204483033,
|
|
"epoch": 0.0276977310242565,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.641,
|
|
"mean_token_accuracy": 0.10911367908120155,
|
|
"num_tokens": 909367.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.214114952087402,
|
|
"epoch": 0.027977506085107573,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.7148,
|
|
"mean_token_accuracy": 0.09811243787407875,
|
|
"num_tokens": 919180.0,
|
|
"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": 202662796492800.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|