Denis641/BiCodeGen-MNTP-CodeSearchNet
05
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 2.928870292887029,5 "eval_steps": 100,6 "global_step": 5600,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.05230125523012552,13 "eval_accuracy": 0.10215735262383098,14 "eval_loss": 5.751355171203613,15 "eval_runtime": 184.0542,16 "eval_samples_per_second": 80.194,17 "eval_steps_per_second": 0.63,18 "step": 10019 },20 {21 "epoch": 0.10460251046025104,22 "eval_accuracy": 0.12205444695787844,23 "eval_loss": 5.327404975891113,24 "eval_runtime": 185.0066,25 "eval_samples_per_second": 79.781,26 "eval_steps_per_second": 0.627,27 "step": 20028 },29 {30 "epoch": 0.15690376569037656,31 "eval_accuracy": 0.13593891242440487,32 "eval_loss": 5.167089939117432,33 "eval_runtime": 184.2672,34 "eval_samples_per_second": 80.101,35 "eval_steps_per_second": 0.63,36 "step": 30037 },38 {39 "epoch": 0.20920502092050208,40 "eval_accuracy": 0.14769799376050313,41 "eval_loss": 5.03983736038208,42 "eval_runtime": 184.4381,43 "eval_samples_per_second": 80.027,44 "eval_steps_per_second": 0.629,45 "step": 40046 },47 {48 "epoch": 0.2615062761506276,49 "grad_norm": 1.4375,50 "learning_rate": 4.564156206415621e-05,51 "loss": 5.3792,52 "step": 50053 },54 {55 "epoch": 0.2615062761506276,56 "eval_accuracy": 0.15595240798761018,57 "eval_loss": 4.976524353027344,58 "eval_runtime": 184.12,59 "eval_samples_per_second": 80.165,60 "eval_steps_per_second": 0.63,61 "step": 50062 },63 {64 "epoch": 0.3138075313807531,65 "eval_accuracy": 0.16177388065723808,66 "eval_loss": 4.927117347717285,67 "eval_runtime": 184.3296,68 "eval_samples_per_second": 80.074,69 "eval_steps_per_second": 0.629,70 "step": 60071 },72 {73 "epoch": 0.36610878661087864,74 "eval_accuracy": 0.16636529077842815,75 "eval_loss": 4.898035049438477,76 "eval_runtime": 184.3142,77 "eval_samples_per_second": 80.081,78 "eval_steps_per_second": 0.629,79 "step": 70080 },81 {82 "epoch": 0.41841004184100417,83 "eval_accuracy": 0.1694151293866953,84 "eval_loss": 4.875,85 "eval_runtime": 185.3011,86 "eval_samples_per_second": 79.654,87 "eval_steps_per_second": 0.626,88 "step": 80089 },90 {91 "epoch": 0.4707112970711297,92 "eval_accuracy": 0.17219956637036357,93 "eval_loss": 4.854589939117432,94 "eval_runtime": 184.8498,95 "eval_samples_per_second": 79.849,96 "eval_steps_per_second": 0.628,97 "step": 90098 },99 {100 "epoch": 0.5230125523012552,101 "grad_norm": 1.2265625,102 "learning_rate": 4.128312412831242e-05,103 "loss": 4.8385,104 "step": 1000105 },106 {107 "epoch": 0.5230125523012552,108 "eval_accuracy": 0.17474116808932638,109 "eval_loss": 4.833265781402588,110 "eval_runtime": 184.5693,111 "eval_samples_per_second": 79.97,112 "eval_steps_per_second": 0.628,113 "step": 1000114 },115 {116 "epoch": 0.5753138075313807,117 "eval_accuracy": 0.17643840085242252,118 "eval_loss": 4.817920207977295,119 "eval_runtime": 184.1405,120 "eval_samples_per_second": 80.156,121 "eval_steps_per_second": 0.63,122 "step": 1100123 },124 {125 "epoch": 0.6276150627615062,126 "eval_accuracy": 0.1773740269532399,127 "eval_loss": 4.811416149139404,128 "eval_runtime": 184.0435,129 "eval_samples_per_second": 80.198,130 "eval_steps_per_second": 0.63,131 "step": 1200132 },133 {134 "epoch": 0.6799163179916318,135 "eval_accuracy": 0.17846844188652472,136 "eval_loss": 4.802201747894287,137 "eval_runtime": 185.0497,138 "eval_samples_per_second": 79.762,139 "eval_steps_per_second": 0.627,140 "step": 1300141 },142 {143 "epoch": 0.7322175732217573,144 "eval_accuracy": 0.1789691379773809,145 "eval_loss": 4.7901930809021,146 "eval_runtime": 184.9855,147 "eval_samples_per_second": 79.79,148 "eval_steps_per_second": 0.627,149 "step": 1400150 },151 {152 "epoch": 0.7845188284518828,153 "grad_norm": 1.5703125,154 "learning_rate": 3.6924686192468624e-05,155 "loss": 4.7486,156 "step": 1500157 },158 {159 "epoch": 0.7845188284518828,160 "eval_accuracy": 0.18001693513769368,161 "eval_loss": 4.785585880279541,162 "eval_runtime": 183.9922,163 "eval_samples_per_second": 80.221,164 "eval_steps_per_second": 0.63,165 "step": 1500166 },167 {168 "epoch": 0.8368200836820083,169 "eval_accuracy": 0.18058760065826654,170 "eval_loss": 4.779539108276367,171 "eval_runtime": 184.1103,172 "eval_samples_per_second": 80.169,173 "eval_steps_per_second": 0.63,174 "step": 1600175 },176 {177 "epoch": 0.8891213389121339,178 "eval_accuracy": 0.18107188047030814,179 "eval_loss": 4.776101112365723,180 "eval_runtime": 184.1674,181 "eval_samples_per_second": 80.144,182 "eval_steps_per_second": 0.63,183 "step": 1700184 },185 {186 "epoch": 0.9414225941422594,187 "eval_accuracy": 0.18138128706525738,188 "eval_loss": 4.7754740715026855,189 "eval_runtime": 184.197,190 "eval_samples_per_second": 80.132,191 "eval_steps_per_second": 0.63,192 "step": 1800193 },194 {195 "epoch": 0.9937238493723849,196 "eval_accuracy": 0.18191716908996425,197 "eval_loss": 4.767343997955322,198 "eval_runtime": 184.0497,199 "eval_samples_per_second": 80.196,200 "eval_steps_per_second": 0.63,201 "step": 1900202 },203 {204 "epoch": 1.0460251046025104,205 "grad_norm": 1.203125,206 "learning_rate": 3.2566248256624825e-05,207 "loss": 4.7159,208 "step": 2000209 },210 {211 "epoch": 1.0460251046025104,212 "eval_accuracy": 0.18160189904678178,213 "eval_loss": 4.769783020019531,214 "eval_runtime": 185.2615,215 "eval_samples_per_second": 79.671,216 "eval_steps_per_second": 0.626,217 "step": 2000218 },219 {220 "epoch": 1.098326359832636,221 "eval_accuracy": 0.18223365637995678,222 "eval_loss": 4.763906002044678,223 "eval_runtime": 184.3991,224 "eval_samples_per_second": 80.044,225 "eval_steps_per_second": 0.629,226 "step": 2100227 },228 {229 "epoch": 1.1506276150627615,230 "eval_accuracy": 0.18256457647135646,231 "eval_loss": 4.761280536651611,232 "eval_runtime": 185.0808,233 "eval_samples_per_second": 79.749,234 "eval_steps_per_second": 0.627,235 "step": 2200236 },237 {238 "epoch": 1.202928870292887,239 "eval_accuracy": 0.18278985816542134,240 "eval_loss": 4.7557759284973145,241 "eval_runtime": 184.6355,242 "eval_samples_per_second": 79.941,243 "eval_steps_per_second": 0.628,244 "step": 2300245 },246 {247 "epoch": 1.2552301255230125,248 "eval_accuracy": 0.18294602847055624,249 "eval_loss": 4.75867223739624,250 "eval_runtime": 184.9626,251 "eval_samples_per_second": 79.8,252 "eval_steps_per_second": 0.627,253 "step": 2400254 },255 {256 "epoch": 1.3075313807531381,257 "grad_norm": 1.75,258 "learning_rate": 2.8207810320781032e-05,259 "loss": 4.6997,260 "step": 2500261 },262 {263 "epoch": 1.3075313807531381,264 "eval_accuracy": 0.18342178332552747,265 "eval_loss": 4.754149913787842,266 "eval_runtime": 185.1377,267 "eval_samples_per_second": 79.724,268 "eval_steps_per_second": 0.627,269 "step": 2500270 },271 {272 "epoch": 1.3598326359832635,273 "eval_accuracy": 0.1834138543952502,274 "eval_loss": 4.75181245803833,275 "eval_runtime": 184.6159,276 "eval_samples_per_second": 79.95,277 "eval_steps_per_second": 0.628,278 "step": 2600279 },280 {281 "epoch": 1.4121338912133892,282 "eval_accuracy": 0.1836629457159783,283 "eval_loss": 4.746477127075195,284 "eval_runtime": 185.4196,285 "eval_samples_per_second": 79.603,286 "eval_steps_per_second": 0.626,287 "step": 2700288 },289 {290 "epoch": 1.4644351464435146,291 "eval_accuracy": 0.18383634667304455,292 "eval_loss": 4.750728130340576,293 "eval_runtime": 184.604,294 "eval_samples_per_second": 79.955,295 "eval_steps_per_second": 0.628,296 "step": 2800297 },298 {299 "epoch": 1.5167364016736402,300 "eval_accuracy": 0.1834985251891202,301 "eval_loss": 4.751083850860596,302 "eval_runtime": 183.9829,303 "eval_samples_per_second": 80.225,304 "eval_steps_per_second": 0.63,305 "step": 2900306 },307 {308 "epoch": 1.5690376569037658,309 "grad_norm": 1.1328125,310 "learning_rate": 2.3849372384937242e-05,311 "loss": 4.6905,312 "step": 3000313 },314 {315 "epoch": 1.5690376569037658,316 "eval_accuracy": 0.18385013375825832,317 "eval_loss": 4.750813007354736,318 "eval_runtime": 184.3904,319 "eval_samples_per_second": 80.048,320 "eval_steps_per_second": 0.629,321 "step": 3000322 },323 {324 "epoch": 1.6213389121338913,325 "eval_accuracy": 0.18415784707342428,326 "eval_loss": 4.746849536895752,327 "eval_runtime": 184.2103,328 "eval_samples_per_second": 80.126,329 "eval_steps_per_second": 0.63,330 "step": 3100331 },332 {333 "epoch": 1.6736401673640167,334 "eval_accuracy": 0.1842263748450887,335 "eval_loss": 4.746747970581055,336 "eval_runtime": 184.1204,337 "eval_samples_per_second": 80.165,338 "eval_steps_per_second": 0.63,339 "step": 3200340 },341 {342 "epoch": 1.7259414225941423,343 "eval_accuracy": 0.18430792521293346,344 "eval_loss": 4.745037078857422,345 "eval_runtime": 184.2802,346 "eval_samples_per_second": 80.095,347 "eval_steps_per_second": 0.629,348 "step": 3300349 },350 {351 "epoch": 1.778242677824268,352 "eval_accuracy": 0.184430838994751,353 "eval_loss": 4.746375560760498,354 "eval_runtime": 184.234,355 "eval_samples_per_second": 80.116,356 "eval_steps_per_second": 0.63,357 "step": 3400358 },359 {360 "epoch": 1.8305439330543933,361 "grad_norm": 2.46875,362 "learning_rate": 1.9490934449093446e-05,363 "loss": 4.687,364 "step": 3500365 },366 {367 "epoch": 1.8305439330543933,368 "eval_accuracy": 0.18449599939000436,369 "eval_loss": 4.7423272132873535,370 "eval_runtime": 184.1596,371 "eval_samples_per_second": 80.148,372 "eval_steps_per_second": 0.63,373 "step": 3500374 },375 {376 "epoch": 1.8828451882845187,377 "eval_accuracy": 0.18466320271445863,378 "eval_loss": 4.74322509765625,379 "eval_runtime": 183.9907,380 "eval_samples_per_second": 80.221,381 "eval_steps_per_second": 0.63,382 "step": 3600383 },384 {385 "epoch": 1.9351464435146444,386 "eval_accuracy": 0.18432753476168612,387 "eval_loss": 4.744410514831543,388 "eval_runtime": 183.9473,389 "eval_samples_per_second": 80.24,390 "eval_steps_per_second": 0.631,391 "step": 3700392 },393 {394 "epoch": 1.98744769874477,395 "eval_accuracy": 0.18470474137931034,396 "eval_loss": 4.74097204208374,397 "eval_runtime": 184.0059,398 "eval_samples_per_second": 80.215,399 "eval_steps_per_second": 0.63,400 "step": 3800401 },402 {403 "epoch": 2.0397489539748954,404 "eval_accuracy": 0.1845896528685144,405 "eval_loss": 4.744495391845703,406 "eval_runtime": 184.1792,407 "eval_samples_per_second": 80.139,408 "eval_steps_per_second": 0.63,409 "step": 3900410 },411 {412 "epoch": 2.092050209205021,413 "grad_norm": 1.015625,414 "learning_rate": 1.5132496513249652e-05,415 "loss": 4.6822,416 "step": 4000417 },418 {419 "epoch": 2.092050209205021,420 "eval_accuracy": 0.1840621610356637,421 "eval_loss": 4.743766784667969,422 "eval_runtime": 184.0079,423 "eval_samples_per_second": 80.214,424 "eval_steps_per_second": 0.63,425 "step": 4000426 },427 {428 "epoch": 2.1443514644351462,429 "eval_accuracy": 0.18438914008407487,430 "eval_loss": 4.742242336273193,431 "eval_runtime": 184.052,432 "eval_samples_per_second": 80.195,433 "eval_steps_per_second": 0.63,434 "step": 4100435 },436 {437 "epoch": 2.196652719665272,438 "eval_accuracy": 0.18475792298529636,439 "eval_loss": 4.741429328918457,440 "eval_runtime": 184.719,441 "eval_samples_per_second": 79.905,442 "eval_steps_per_second": 0.628,443 "step": 4200444 },445 {446 "epoch": 2.2489539748953975,447 "eval_accuracy": 0.1848885987251326,448 "eval_loss": 4.740514755249023,449 "eval_runtime": 184.0592,450 "eval_samples_per_second": 80.192,451 "eval_steps_per_second": 0.63,452 "step": 4300453 },454 {455 "epoch": 2.301255230125523,456 "eval_accuracy": 0.18497058180948975,457 "eval_loss": 4.738888740539551,458 "eval_runtime": 184.7818,459 "eval_samples_per_second": 79.878,460 "eval_steps_per_second": 0.628,461 "step": 4400462 },463 {464 "epoch": 2.3535564853556483,465 "grad_norm": 1.5546875,466 "learning_rate": 1.0774058577405859e-05,467 "loss": 4.6787,468 "step": 4500469 },470 {471 "epoch": 2.3535564853556483,472 "eval_accuracy": 0.18458716879639453,473 "eval_loss": 4.743495941162109,474 "eval_runtime": 185.012,475 "eval_samples_per_second": 79.779,476 "eval_steps_per_second": 0.627,477 "step": 4500478 },479 {480 "epoch": 2.405857740585774,481 "eval_accuracy": 0.18485314383095383,482 "eval_loss": 4.742412090301514,483 "eval_runtime": 184.018,484 "eval_samples_per_second": 80.21,485 "eval_steps_per_second": 0.63,486 "step": 4600487 },488 {489 "epoch": 2.4581589958158996,490 "eval_accuracy": 0.18451762653729054,491 "eval_loss": 4.744495391845703,492 "eval_runtime": 184.1989,493 "eval_samples_per_second": 80.131,494 "eval_steps_per_second": 0.63,495 "step": 4700496 },497 {498 "epoch": 2.510460251046025,499 "eval_accuracy": 0.18499375944599727,500 "eval_loss": 4.742056369781494,501 "eval_runtime": 184.9544,502 "eval_samples_per_second": 79.803,503 "eval_steps_per_second": 0.627,504 "step": 4800505 },506 {507 "epoch": 2.562761506276151,508 "eval_accuracy": 0.1845635125940072,509 "eval_loss": 4.74495267868042,510 "eval_runtime": 184.45,511 "eval_samples_per_second": 80.022,512 "eval_steps_per_second": 0.629,513 "step": 4900514 },515 {516 "epoch": 2.6150627615062763,517 "grad_norm": 1.25,518 "learning_rate": 6.415620641562065e-06,519 "loss": 4.6809,520 "step": 5000521 },522 {523 "epoch": 2.6150627615062763,524 "eval_accuracy": 0.18461182940869522,525 "eval_loss": 4.739973068237305,526 "eval_runtime": 185.0928,527 "eval_samples_per_second": 79.744,528 "eval_steps_per_second": 0.627,529 "step": 5000530 },531 {532 "epoch": 2.6673640167364017,533 "eval_accuracy": 0.18471778348337312,534 "eval_loss": 4.740243911743164,535 "eval_runtime": 184.3979,536 "eval_samples_per_second": 80.044,537 "eval_steps_per_second": 0.629,538 "step": 5100539 },540 {541 "epoch": 2.719665271966527,542 "eval_accuracy": 0.18489124970131737,543 "eval_loss": 4.738804340362549,544 "eval_runtime": 184.6067,545 "eval_samples_per_second": 79.954,546 "eval_steps_per_second": 0.628,547 "step": 5200548 },549 {550 "epoch": 2.7719665271966525,551 "eval_accuracy": 0.18462476994845917,552 "eval_loss": 4.741514205932617,553 "eval_runtime": 184.2661,554 "eval_samples_per_second": 80.102,555 "eval_steps_per_second": 0.63,556 "step": 5300557 },558 {559 "epoch": 2.8242677824267783,560 "eval_accuracy": 0.18472434773026122,561 "eval_loss": 4.739160060882568,562 "eval_runtime": 185.2276,563 "eval_samples_per_second": 79.686,564 "eval_steps_per_second": 0.626,565 "step": 5400566 },567 {568 "epoch": 2.8765690376569037,569 "grad_norm": 1.6796875,570 "learning_rate": 2.057182705718271e-06,571 "loss": 4.6819,572 "step": 5500573 },574 {575 "epoch": 2.8765690376569037,576 "eval_accuracy": 0.18477532080101736,577 "eval_loss": 4.742056369781494,578 "eval_runtime": 184.4871,579 "eval_samples_per_second": 80.006,580 "eval_steps_per_second": 0.629,581 "step": 5500582 },583 {584 "epoch": 2.928870292887029,585 "eval_accuracy": 0.18478108276943064,586 "eval_loss": 4.740701198577881,587 "eval_runtime": 184.224,588 "eval_samples_per_second": 80.12,589 "eval_steps_per_second": 0.63,590 "step": 5600591 }592 ],593 "logging_steps": 500,594 "max_steps": 5736,595 "num_input_tokens_seen": 0,596 "num_train_epochs": 3,597 "save_steps": 200,598 "total_flos": 6.743272090868122e+17,599 "train_batch_size": 128,600 "trial_name": null,601 "trial_params": null602}603