maximxls/text-normalization-ru-terrible
489
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 21.650061500615006,5 "global_step": 316827,6 "is_hyper_param_search": false,7 "is_local_process_zero": true,8 "is_world_process_zero": true,9 "log_history": [10 {11 "epoch": 0.0,12 "learning_rate": 2.840909090909091e-06,13 "loss": 4.4773,14 "step": 115 },16 {17 "epoch": 0.02,18 "learning_rate": 0.0005,19 "loss": 4.2748,20 "step": 17621 },22 {23 "epoch": 0.02,24 "eval_loss": 5.831331253051758,25 "eval_max_distance": 201,26 "eval_mean_distance": 116,27 "eval_runtime": 16.9557,28 "eval_samples_per_second": 29.489,29 "eval_steps_per_second": 0.59,30 "step": 17631 },32 {33 "epoch": 0.03,34 "learning_rate": 0.001,35 "loss": 3.2983,36 "step": 35237 },38 {39 "epoch": 0.03,40 "eval_loss": 5.142276287078857,41 "eval_max_distance": 186,42 "eval_mean_distance": 102,43 "eval_runtime": 14.0636,44 "eval_samples_per_second": 35.553,45 "eval_steps_per_second": 0.711,46 "step": 35247 },48 {49 "epoch": 0.05,50 "learning_rate": 0.0009949384562291498,51 "loss": 2.9967,52 "step": 52853 },54 {55 "epoch": 0.05,56 "eval_loss": 4.896122932434082,57 "eval_max_distance": 196,58 "eval_mean_distance": 104,59 "eval_runtime": 15.2018,60 "eval_samples_per_second": 32.891,61 "eval_steps_per_second": 0.658,62 "step": 52863 },64 {65 "epoch": 0.06,66 "learning_rate": 0.0009898769124582999,67 "loss": 2.9064,68 "step": 70469 },70 {71 "epoch": 0.06,72 "eval_loss": 4.704981803894043,73 "eval_max_distance": 171,74 "eval_mean_distance": 99,75 "eval_runtime": 14.8835,76 "eval_samples_per_second": 33.594,77 "eval_steps_per_second": 0.672,78 "step": 70479 },80 {81 "epoch": 0.08,82 "learning_rate": 0.0009848153686874497,83 "loss": 2.8444,84 "step": 88085 },86 {87 "epoch": 0.08,88 "eval_loss": 4.648359775543213,89 "eval_max_distance": 173,90 "eval_mean_distance": 99,91 "eval_runtime": 13.5358,92 "eval_samples_per_second": 36.939,93 "eval_steps_per_second": 0.739,94 "step": 88095 },96 {97 "epoch": 0.09,98 "learning_rate": 0.0009797538249165995,99 "loss": 2.7359,100 "step": 1056101 },102 {103 "epoch": 0.09,104 "eval_loss": 4.639862537384033,105 "eval_max_distance": 200,106 "eval_mean_distance": 96,107 "eval_runtime": 13.4266,108 "eval_samples_per_second": 37.239,109 "eval_steps_per_second": 0.745,110 "step": 1056111 },112 {113 "epoch": 0.11,114 "learning_rate": 0.0009746922811457495,115 "loss": 2.667,116 "step": 1232117 },118 {119 "epoch": 0.11,120 "eval_loss": 4.517767906188965,121 "eval_max_distance": 160,122 "eval_mean_distance": 93,123 "eval_runtime": 13.7165,124 "eval_samples_per_second": 36.453,125 "eval_steps_per_second": 0.729,126 "step": 1232127 },128 {129 "epoch": 0.12,130 "learning_rate": 0.0009696307373748994,131 "loss": 2.6454,132 "step": 1408133 },134 {135 "epoch": 0.12,136 "eval_loss": 4.468171119689941,137 "eval_max_distance": 154,138 "eval_mean_distance": 92,139 "eval_runtime": 14.5339,140 "eval_samples_per_second": 34.402,141 "eval_steps_per_second": 0.688,142 "step": 1408143 },144 {145 "epoch": 0.14,146 "learning_rate": 0.0009645691936040492,147 "loss": 2.5737,148 "step": 1584149 },150 {151 "epoch": 0.14,152 "eval_loss": 4.346231460571289,153 "eval_max_distance": 151,154 "eval_mean_distance": 91,155 "eval_runtime": 13.9596,156 "eval_samples_per_second": 35.818,157 "eval_steps_per_second": 0.716,158 "step": 1584159 },160 {161 "epoch": 0.15,162 "learning_rate": 0.0009595076498331991,163 "loss": 2.5569,164 "step": 1760165 },166 {167 "epoch": 0.15,168 "eval_loss": 4.252463340759277,169 "eval_max_distance": 166,170 "eval_mean_distance": 92,171 "eval_runtime": 13.7296,172 "eval_samples_per_second": 36.418,173 "eval_steps_per_second": 0.728,174 "step": 1760175 },176 {177 "epoch": 0.17,178 "learning_rate": 0.0009544461060623491,179 "loss": 2.5114,180 "step": 1936181 },182 {183 "epoch": 0.17,184 "eval_loss": 4.216597557067871,185 "eval_max_distance": 164,186 "eval_mean_distance": 90,187 "eval_runtime": 14.6796,188 "eval_samples_per_second": 34.061,189 "eval_steps_per_second": 0.681,190 "step": 1936191 },192 {193 "epoch": 0.18,194 "learning_rate": 0.0009493845622914989,195 "loss": 2.4448,196 "step": 2112197 },198 {199 "epoch": 0.18,200 "eval_loss": 4.229903697967529,201 "eval_max_distance": 151,202 "eval_mean_distance": 90,203 "eval_runtime": 14.5473,204 "eval_samples_per_second": 34.371,205 "eval_steps_per_second": 0.687,206 "step": 2112207 },208 {209 "epoch": 0.2,210 "learning_rate": 0.0009443230185206488,211 "loss": 2.3847,212 "step": 2288213 },214 {215 "epoch": 0.2,216 "eval_loss": 4.108386039733887,217 "eval_max_distance": 146,218 "eval_mean_distance": 90,219 "eval_runtime": 14.7702,220 "eval_samples_per_second": 33.852,221 "eval_steps_per_second": 0.677,222 "step": 2288223 },224 {225 "epoch": 0.21,226 "learning_rate": 0.0009392614747497988,227 "loss": 2.1415,228 "step": 2464229 },230 {231 "epoch": 0.21,232 "eval_loss": 3.6208255290985107,233 "eval_max_distance": 130,234 "eval_mean_distance": 78,235 "eval_runtime": 15.5259,236 "eval_samples_per_second": 32.204,237 "eval_steps_per_second": 0.644,238 "step": 2464239 },240 {241 "epoch": 0.23,242 "learning_rate": 0.0009341999309789486,243 "loss": 1.0359,244 "step": 2640245 },246 {247 "epoch": 0.23,248 "eval_loss": 1.5739717483520508,249 "eval_max_distance": 139,250 "eval_mean_distance": 41,251 "eval_runtime": 17.5626,252 "eval_samples_per_second": 28.47,253 "eval_steps_per_second": 0.569,254 "step": 2640255 },256 {257 "epoch": 0.24,258 "learning_rate": 0.0009291383872080985,259 "loss": 0.5688,260 "step": 2816261 },262 {263 "epoch": 0.24,264 "eval_loss": 1.293850064277649,265 "eval_max_distance": 127,266 "eval_mean_distance": 35,267 "eval_runtime": 13.8285,268 "eval_samples_per_second": 36.157,269 "eval_steps_per_second": 0.723,270 "step": 2816271 },272 {273 "epoch": 0.26,274 "learning_rate": 0.0009240768434372484,275 "loss": 0.5115,276 "step": 2992277 },278 {279 "epoch": 0.26,280 "eval_loss": 1.188537836074829,281 "eval_max_distance": 120,282 "eval_mean_distance": 32,283 "eval_runtime": 15.477,284 "eval_samples_per_second": 32.306,285 "eval_steps_per_second": 0.646,286 "step": 2992287 },288 {289 "epoch": 0.27,290 "learning_rate": 0.0009190152996663983,291 "loss": 0.4662,292 "step": 3168293 },294 {295 "epoch": 0.27,296 "eval_loss": 1.1915712356567383,297 "eval_max_distance": 127,298 "eval_mean_distance": 32,299 "eval_runtime": 16.2261,300 "eval_samples_per_second": 30.815,301 "eval_steps_per_second": 0.616,302 "step": 3168303 },304 {305 "epoch": 0.29,306 "learning_rate": 0.0009139537558955482,307 "loss": 0.4401,308 "step": 3344309 },310 {311 "epoch": 0.29,312 "eval_loss": 1.15479576587677,313 "eval_max_distance": 126,314 "eval_mean_distance": 31,315 "eval_runtime": 16.6846,316 "eval_samples_per_second": 29.968,317 "eval_steps_per_second": 0.599,318 "step": 3344319 },320 {321 "epoch": 0.3,322 "learning_rate": 0.0009088922121246981,323 "loss": 0.408,324 "step": 3520325 },326 {327 "epoch": 0.3,328 "eval_loss": 1.1046110391616821,329 "eval_max_distance": 103,330 "eval_mean_distance": 29,331 "eval_runtime": 14.9865,332 "eval_samples_per_second": 33.363,333 "eval_steps_per_second": 0.667,334 "step": 3520335 },336 {337 "epoch": 0.32,338 "learning_rate": 0.000903830668353848,339 "loss": 0.3858,340 "step": 3696341 },342 {343 "epoch": 0.32,344 "eval_loss": 0.9291953444480896,345 "eval_max_distance": 135,346 "eval_mean_distance": 26,347 "eval_runtime": 13.5564,348 "eval_samples_per_second": 36.883,349 "eval_steps_per_second": 0.738,350 "step": 3696351 },352 {353 "epoch": 0.33,354 "learning_rate": 0.0008987691245829979,355 "loss": 0.3586,356 "step": 3872357 },358 {359 "epoch": 0.33,360 "eval_loss": 0.7040215730667114,361 "eval_max_distance": 115,362 "eval_mean_distance": 21,363 "eval_runtime": 15.3203,364 "eval_samples_per_second": 32.636,365 "eval_steps_per_second": 0.653,366 "step": 3872367 },368 {369 "epoch": 0.35,370 "learning_rate": 0.0008937075808121477,371 "loss": 0.3237,372 "step": 4048373 },374 {375 "epoch": 0.35,376 "eval_loss": 0.5791512131690979,377 "eval_max_distance": 139,378 "eval_mean_distance": 17,379 "eval_runtime": 14.8657,380 "eval_samples_per_second": 33.634,381 "eval_steps_per_second": 0.673,382 "step": 4048383 },384 {385 "epoch": 0.36,386 "learning_rate": 0.0008886460370412976,387 "loss": 0.2783,388 "step": 4224389 },390 {391 "epoch": 0.36,392 "eval_loss": 0.4473143219947815,393 "eval_max_distance": 122,394 "eval_mean_distance": 14,395 "eval_runtime": 13.1694,396 "eval_samples_per_second": 37.967,397 "eval_steps_per_second": 0.759,398 "step": 4224399 },400 {401 "epoch": 0.38,402 "learning_rate": 0.0008835844932704476,403 "loss": 0.2666,404 "step": 4400405 },406 {407 "epoch": 0.38,408 "eval_loss": 0.3882352411746979,409 "eval_max_distance": 125,410 "eval_mean_distance": 11,411 "eval_runtime": 14.0887,412 "eval_samples_per_second": 35.489,413 "eval_steps_per_second": 0.71,414 "step": 4400415 },416 {417 "epoch": 0.39,418 "learning_rate": 0.0008785229494995974,419 "loss": 0.249,420 "step": 4576421 },422 {423 "epoch": 0.39,424 "eval_loss": 0.3606760799884796,425 "eval_max_distance": 126,426 "eval_mean_distance": 11,427 "eval_runtime": 13.741,428 "eval_samples_per_second": 36.388,429 "eval_steps_per_second": 0.728,430 "step": 4576431 },432 {433 "epoch": 0.41,434 "learning_rate": 0.0008734614057287473,435 "loss": 0.2371,436 "step": 4752437 },438 {439 "epoch": 0.41,440 "eval_loss": 0.3074778616428375,441 "eval_max_distance": 111,442 "eval_mean_distance": 9,443 "eval_runtime": 13.0071,444 "eval_samples_per_second": 38.441,445 "eval_steps_per_second": 0.769,446 "step": 4752447 },448 {449 "epoch": 0.42,450 "learning_rate": 0.0008683998619578973,451 "loss": 0.2231,452 "step": 4928453 },454 {455 "epoch": 0.42,456 "eval_loss": 0.2844279706478119,457 "eval_max_distance": 123,458 "eval_mean_distance": 8,459 "eval_runtime": 12.8112,460 "eval_samples_per_second": 39.028,461 "eval_steps_per_second": 0.781,462 "step": 4928463 },464 {465 "epoch": 0.44,466 "learning_rate": 0.0008633383181870471,467 "loss": 0.2114,468 "step": 5104469 },470 {471 "epoch": 0.44,472 "eval_loss": 0.27491652965545654,473 "eval_max_distance": 133,474 "eval_mean_distance": 8,475 "eval_runtime": 11.3043,476 "eval_samples_per_second": 44.231,477 "eval_steps_per_second": 0.885,478 "step": 5104479 },480 {481 "epoch": 0.45,482 "learning_rate": 0.000858276774416197,483 "loss": 0.1946,484 "step": 5280485 },486 {487 "epoch": 0.45,488 "eval_loss": 0.2683752775192261,489 "eval_max_distance": 124,490 "eval_mean_distance": 8,491 "eval_runtime": 11.6576,492 "eval_samples_per_second": 42.891,493 "eval_steps_per_second": 0.858,494 "step": 5280495 },496 {497 "epoch": 0.47,498 "learning_rate": 0.000853215230645347,499 "loss": 0.1893,500 "step": 5456501 },502 {503 "epoch": 0.47,504 "eval_loss": 0.25331711769104004,505 "eval_max_distance": 129,506 "eval_mean_distance": 7,507 "eval_runtime": 11.955,508 "eval_samples_per_second": 41.824,509 "eval_steps_per_second": 0.836,510 "step": 5456511 },512 {513 "epoch": 0.48,514 "learning_rate": 0.0008481536868744968,515 "loss": 0.1865,516 "step": 5632517 },518 {519 "epoch": 0.48,520 "eval_loss": 0.22615253925323486,521 "eval_max_distance": 129,522 "eval_mean_distance": 7,523 "eval_runtime": 10.581,524 "eval_samples_per_second": 47.255,525 "eval_steps_per_second": 0.945,526 "step": 5632527 },528 {529 "epoch": 0.5,530 "learning_rate": 0.0008430921431036467,531 "loss": 0.1769,532 "step": 5808533 },534 {535 "epoch": 0.5,536 "eval_loss": 0.22688022255897522,537 "eval_max_distance": 137,538 "eval_mean_distance": 6,539 "eval_runtime": 10.5367,540 "eval_samples_per_second": 47.453,541 "eval_steps_per_second": 0.949,542 "step": 5808543 },544 {545 "epoch": 0.51,546 "learning_rate": 0.0008380305993327965,547 "loss": 0.172,548 "step": 5984549 },550 {551 "epoch": 0.51,552 "eval_loss": 0.21938543021678925,553 "eval_max_distance": 129,554 "eval_mean_distance": 6,555 "eval_runtime": 10.9867,556 "eval_samples_per_second": 45.51,557 "eval_steps_per_second": 0.91,558 "step": 5984559 },560 {561 "epoch": 0.53,562 "learning_rate": 0.0008329690555619465,563 "loss": 0.1594,564 "step": 6160565 },566 {567 "epoch": 0.53,568 "eval_loss": 0.20960509777069092,569 "eval_max_distance": 138,570 "eval_mean_distance": 6,571 "eval_runtime": 8.8034,572 "eval_samples_per_second": 56.796,573 "eval_steps_per_second": 1.136,574 "step": 6160575 },576 {577 "epoch": 0.54,578 "learning_rate": 0.0008279075117910963,579 "loss": 0.158,580 "step": 6336581 },582 {583 "epoch": 0.54,584 "eval_loss": 0.20247827470302582,585 "eval_max_distance": 122,586 "eval_mean_distance": 6,587 "eval_runtime": 9.7934,588 "eval_samples_per_second": 51.055,589 "eval_steps_per_second": 1.021,590 "step": 6336591 },592 {593 "epoch": 0.56,594 "learning_rate": 0.0008228459680202461,595 "loss": 0.1436,596 "step": 6512597 },598 {599 "epoch": 0.56,600 "eval_loss": 0.20747961103916168,601 "eval_max_distance": 128,602 "eval_mean_distance": 6,603 "eval_runtime": 10.4315,604 "eval_samples_per_second": 47.932,605 "eval_steps_per_second": 0.959,606 "step": 6512607 },608 {609 "epoch": 0.57,610 "learning_rate": 0.000817784424249396,611 "loss": 0.1398,612 "step": 6688613 },614 {615 "epoch": 0.57,616 "eval_loss": 0.19245657324790955,617 "eval_max_distance": 128,618 "eval_mean_distance": 6,619 "eval_runtime": 9.4807,620 "eval_samples_per_second": 52.739,621 "eval_steps_per_second": 1.055,622 "step": 6688623 },624 {625 "epoch": 0.59,626 "learning_rate": 0.000812722880478546,627 "loss": 0.1326,628 "step": 6864629 },630 {631 "epoch": 0.59,632 "eval_loss": 0.1919807493686676,633 "eval_max_distance": 125,634 "eval_mean_distance": 6,635 "eval_runtime": 9.3305,636 "eval_samples_per_second": 53.588,637 "eval_steps_per_second": 1.072,638 "step": 6864639 },640 {641 "epoch": 0.6,642 "learning_rate": 0.0008076613367076958,643 "loss": 0.1321,644 "step": 7040645 },646 {647 "epoch": 0.6,648 "eval_loss": 0.18875986337661743,649 "eval_max_distance": 135,650 "eval_mean_distance": 5,651 "eval_runtime": 7.8177,652 "eval_samples_per_second": 63.958,653 "eval_steps_per_second": 1.279,654 "step": 7040655 },656 {657 "epoch": 0.62,658 "learning_rate": 0.0008025997929368457,659 "loss": 0.1255,660 "step": 7216661 },662 {663 "epoch": 0.62,664 "eval_loss": 0.18568550050258636,665 "eval_max_distance": 144,666 "eval_mean_distance": 6,667 "eval_runtime": 9.4575,668 "eval_samples_per_second": 52.868,669 "eval_steps_per_second": 1.057,670 "step": 7216671 },672 {673 "epoch": 0.63,674 "learning_rate": 0.0007975382491659956,675 "loss": 0.1257,676 "step": 7392677 },678 {679 "epoch": 0.63,680 "eval_loss": 0.18572939932346344,681 "eval_max_distance": 113,682 "eval_mean_distance": 5,683 "eval_runtime": 7.6787,684 "eval_samples_per_second": 65.115,685 "eval_steps_per_second": 1.302,686 "step": 7392687 },688 {689 "epoch": 0.65,690 "learning_rate": 0.0007924767053951455,691 "loss": 0.1175,692 "step": 7568693 },694 {695 "epoch": 0.65,696 "eval_loss": 0.17028774321079254,697 "eval_max_distance": 124,698 "eval_mean_distance": 5,699 "eval_runtime": 7.3771,700 "eval_samples_per_second": 67.777,701 "eval_steps_per_second": 1.356,702 "step": 7568703 },704 {705 "epoch": 0.66,706 "learning_rate": 0.0007874151616242954,707 "loss": 0.1109,708 "step": 7744709 },710 {711 "epoch": 0.66,712 "eval_loss": 0.1755068004131317,713 "eval_max_distance": 126,714 "eval_mean_distance": 5,715 "eval_runtime": 7.0941,716 "eval_samples_per_second": 70.482,717 "eval_steps_per_second": 1.41,718 "step": 7744719 },720 {721 "epoch": 0.68,722 "learning_rate": 0.0007823536178534452,723 "loss": 0.1065,724 "step": 7920725 },726 {727 "epoch": 0.68,728 "eval_loss": 0.17363803088665009,729 "eval_max_distance": 128,730 "eval_mean_distance": 5,731 "eval_runtime": 8.9523,732 "eval_samples_per_second": 55.852,733 "eval_steps_per_second": 1.117,734 "step": 7920735 },736 {737 "epoch": 0.69,738 "learning_rate": 0.0007772920740825952,739 "loss": 0.104,740 "step": 8096741 },742 {743 "epoch": 0.69,744 "eval_loss": 0.16887737810611725,745 "eval_max_distance": 129,746 "eval_mean_distance": 5,747 "eval_runtime": 8.6778,748 "eval_samples_per_second": 57.619,749 "eval_steps_per_second": 1.152,750 "step": 8096751 },752 {753 "epoch": 0.71,754 "learning_rate": 0.0007722305303117451,755 "loss": 0.0997,756 "step": 8272757 },758 {759 "epoch": 0.71,760 "eval_loss": 0.16253332793712616,761 "eval_max_distance": 117,762 "eval_mean_distance": 5,763 "eval_runtime": 9.3982,764 "eval_samples_per_second": 53.202,765 "eval_steps_per_second": 1.064,766 "step": 8272767 },768 {769 "epoch": 0.72,770 "learning_rate": 0.0007671689865408949,771 "loss": 0.0993,772 "step": 8448773 },774 {775 "epoch": 0.72,776 "eval_loss": 0.16519272327423096,777 "eval_max_distance": 120,778 "eval_mean_distance": 5,779 "eval_runtime": 8.3081,780 "eval_samples_per_second": 60.182,781 "eval_steps_per_second": 1.204,782 "step": 8448783 },784 {785 "epoch": 0.74,786 "learning_rate": 0.0007621074427700448,787 "loss": 0.0962,788 "step": 8624789 },790 {791 "epoch": 0.74,792 "eval_loss": 0.16153652966022491,793 "eval_max_distance": 113,794 "eval_mean_distance": 5,795 "eval_runtime": 8.3339,796 "eval_samples_per_second": 59.996,797 "eval_steps_per_second": 1.2,798 "step": 8624799 },800 {801 "epoch": 0.75,802 "learning_rate": 0.0007570458989991948,803 "loss": 0.0904,804 "step": 8800805 },806 {807 "epoch": 0.75,808 "eval_loss": 0.1528134047985077,809 "eval_max_distance": 131,810 "eval_mean_distance": 4,811 "eval_runtime": 8.605,812 "eval_samples_per_second": 58.106,813 "eval_steps_per_second": 1.162,814 "step": 8800815 },816 {817 "epoch": 0.77,818 "learning_rate": 0.0007519843552283446,819 "loss": 0.0873,820 "step": 8976821 },822 {823 "epoch": 0.77,824 "eval_loss": 0.15612851083278656,825 "eval_max_distance": 106,826 "eval_mean_distance": 4,827 "eval_runtime": 9.222,828 "eval_samples_per_second": 54.218,829 "eval_steps_per_second": 1.084,830 "step": 8976831 },832 {833 "epoch": 0.78,834 "learning_rate": 0.0007469228114574945,835 "loss": 0.0858,836 "step": 9152837 },838 {839 "epoch": 0.78,840 "eval_loss": 0.15163549780845642,841 "eval_max_distance": 137,842 "eval_mean_distance": 4,843 "eval_runtime": 8.6808,844 "eval_samples_per_second": 57.598,845 "eval_steps_per_second": 1.152,846 "step": 9152847 },848 {849 "epoch": 0.8,850 "learning_rate": 0.0007418612676866445,851 "loss": 0.082,852 "step": 9328853 },854 {855 "epoch": 0.8,856 "eval_loss": 0.1514330506324768,857 "eval_max_distance": 136,858 "eval_mean_distance": 4,859 "eval_runtime": 8.6904,860 "eval_samples_per_second": 57.535,861 "eval_steps_per_second": 1.151,862 "step": 9328863 },864 {865 "epoch": 0.81,866 "learning_rate": 0.0007367997239157943,867 "loss": 0.0766,868 "step": 9504869 },870 {871 "epoch": 0.81,872 "eval_loss": 0.15236127376556396,873 "eval_max_distance": 148,874 "eval_mean_distance": 5,875 "eval_runtime": 8.5887,876 "eval_samples_per_second": 58.216,877 "eval_steps_per_second": 1.164,878 "step": 9504879 },880 {881 "epoch": 0.83,882 "learning_rate": 0.0007317381801449442,883 "loss": 0.075,884 "step": 9680885 },886 {887 "epoch": 0.83,888 "eval_loss": 0.14720940589904785,889 "eval_max_distance": 138,890 "eval_mean_distance": 4,891 "eval_runtime": 8.12,892 "eval_samples_per_second": 61.576,893 "eval_steps_per_second": 1.232,894 "step": 9680895 },896 {897 "epoch": 0.84,898 "learning_rate": 0.000726676636374094,899 "loss": 0.0785,900 "step": 9856901 },902 {903 "epoch": 0.84,904 "eval_loss": 0.15077313780784607,905 "eval_max_distance": 134,906 "eval_mean_distance": 5,907 "eval_runtime": 7.0224,908 "eval_samples_per_second": 71.201,909 "eval_steps_per_second": 1.424,910 "step": 9856911 },912 {913 "epoch": 0.86,914 "learning_rate": 0.000721615092603244,915 "loss": 0.0696,916 "step": 10032917 },918 {919 "epoch": 0.86,920 "eval_loss": 0.1496003121137619,921 "eval_max_distance": 147,922 "eval_mean_distance": 4,923 "eval_runtime": 8.4446,924 "eval_samples_per_second": 59.209,925 "eval_steps_per_second": 1.184,926 "step": 10032927 },928 {929 "epoch": 0.87,930 "learning_rate": 0.0007165535488323939,931 "loss": 0.0697,932 "step": 10208933 },934 {935 "epoch": 0.87,936 "eval_loss": 0.14819692075252533,937 "eval_max_distance": 133,938 "eval_mean_distance": 4,939 "eval_runtime": 7.7516,940 "eval_samples_per_second": 64.503,941 "eval_steps_per_second": 1.29,942 "step": 10208943 },944 {945 "epoch": 0.89,946 "learning_rate": 0.0007114920050615437,947 "loss": 0.0654,948 "step": 10384949 },950 {951 "epoch": 0.89,952 "eval_loss": 0.14708730578422546,953 "eval_max_distance": 109,954 "eval_mean_distance": 4,955 "eval_runtime": 7.7796,956 "eval_samples_per_second": 64.271,957 "eval_steps_per_second": 1.285,958 "step": 10384959 },960 {961 "epoch": 0.9,962 "learning_rate": 0.0007064304612906937,963 "loss": 0.0656,964 "step": 10560965 },966 {967 "epoch": 0.9,968 "eval_loss": 0.14370670914649963,969 "eval_max_distance": 111,970 "eval_mean_distance": 4,971 "eval_runtime": 7.3904,972 "eval_samples_per_second": 67.655,973 "eval_steps_per_second": 1.353,974 "step": 10560975 },976 {977 "epoch": 0.92,978 "learning_rate": 0.0007013689175198436,979 "loss": 0.0617,980 "step": 10736981 },982 {983 "epoch": 0.92,984 "eval_loss": 0.1476200520992279,985 "eval_max_distance": 127,986 "eval_mean_distance": 5,987 "eval_runtime": 8.6231,988 "eval_samples_per_second": 57.984,989 "eval_steps_per_second": 1.16,990 "step": 10736991 },992 {993 "epoch": 0.93,994 "learning_rate": 0.0006963073737489934,995 "loss": 0.0619,996 "step": 10912997 },998 {999 "epoch": 0.93,1000 "eval_loss": 0.14763715863227844,1001 "eval_max_distance": 122,1002 "eval_mean_distance": 4,1003 "eval_runtime": 8.0149,1004 "eval_samples_per_second": 62.384,1005 "eval_steps_per_second": 1.248,1006 "step": 109121007 },1008 {1009 "epoch": 0.95,1010 "learning_rate": 0.0006912458299781433,1011 "loss": 0.0597,1012 "step": 110881013 },1014 {1015 "epoch": 0.95,1016 "eval_loss": 0.1550969034433365,1017 "eval_max_distance": 130,1018 "eval_mean_distance": 5,1019 "eval_runtime": 8.7957,1020 "eval_samples_per_second": 56.846,1021 "eval_steps_per_second": 1.137,1022 "step": 110881023 },1024 {1025 "epoch": 0.96,1026 "learning_rate": 0.0006861842862072933,1027 "loss": 0.0575,1028 "step": 112641029 },1030 {1031 "epoch": 0.96,1032 "eval_loss": 0.15137550234794617,1033 "eval_max_distance": 124,1034 "eval_mean_distance": 5,1035 "eval_runtime": 8.729,1036 "eval_samples_per_second": 57.28,1037 "eval_steps_per_second": 1.146,1038 "step": 112641039 },1040 {1041 "epoch": 0.98,1042 "learning_rate": 0.0006811227424364431,1043 "loss": 0.0575,1044 "step": 114401045 },1046 {1047 "epoch": 0.98,1048 "eval_loss": 0.15129156410694122,1049 "eval_max_distance": 133,1050 "eval_mean_distance": 4,1051 "eval_runtime": 8.3361,1052 "eval_samples_per_second": 59.98,1053 "eval_steps_per_second": 1.2,1054 "step": 114401055 },1056 {1057 "epoch": 0.99,1058 "learning_rate": 0.000676061198665593,1059 "loss": 0.0561,1060 "step": 116161061 },1062 {1063 "epoch": 0.99,1064 "eval_loss": 0.14808788895606995,1065 "eval_max_distance": 135,1066 "eval_mean_distance": 5,1067 "eval_runtime": 8.5181,1068 "eval_samples_per_second": 58.699,1069 "eval_steps_per_second": 1.174,1070 "step": 116161071 },1072 {1073 "epoch": 1.01,1074 "learning_rate": 0.000670999654894743,1075 "loss": 0.2917,1076 "step": 117921077 },1078 {1079 "epoch": 1.01,1080 "eval_loss": 0.11704121530056,1081 "eval_max_distance": 77,1082 "eval_mean_distance": 3,1083 "eval_runtime": 8.4947,1084 "eval_samples_per_second": 58.86,1085 "eval_steps_per_second": 1.177,1086 "step": 117921087 },1088 {1089 "epoch": 1.02,1090 "learning_rate": 0.0006659381111238928,1091 "loss": 0.2267,1092 "step": 119681093 },1094 {1095 "epoch": 1.02,1096 "eval_loss": 0.09986243396997452,1097 "eval_max_distance": 76,1098 "eval_mean_distance": 3,1099 "eval_runtime": 8.3117,1100 "eval_samples_per_second": 60.156,1101 "eval_steps_per_second": 1.203,1102 "step": 119681103 },1104 {1105 "epoch": 1.04,1106 "learning_rate": 0.0006608765673530427,1107 "loss": 0.1755,1108 "step": 121441109 },1110 {1111 "epoch": 1.04,1112 "eval_loss": 0.09780210256576538,1113 "eval_max_distance": 100,1114 "eval_mean_distance": 2,1115 "eval_runtime": 7.2356,1116 "eval_samples_per_second": 69.103,1117 "eval_steps_per_second": 1.382,1118 "step": 121441119 },1120 {1121 "epoch": 1.05,1122 "learning_rate": 0.0006558150235821925,1123 "loss": 0.1476,1124 "step": 123201125 },1126 {1127 "epoch": 1.05,1128 "eval_loss": 0.09662258625030518,1129 "eval_max_distance": 77,1130 "eval_mean_distance": 2,1131 "eval_runtime": 7.9894,1132 "eval_samples_per_second": 62.583,1133 "eval_steps_per_second": 1.252,1134 "step": 123201135 },1136 {1137 "epoch": 1.07,1138 "learning_rate": 0.0006507534798113425,1139 "loss": 0.1283,1140 "step": 124961141 },1142 {1143 "epoch": 1.07,1144 "eval_loss": 0.09604214131832123,1145 "eval_max_distance": 92,1146 "eval_mean_distance": 2,1147 "eval_runtime": 9.1331,1148 "eval_samples_per_second": 54.746,1149 "eval_steps_per_second": 1.095,1150 "step": 124961151 },1152 {1153 "epoch": 1.08,1154 "learning_rate": 0.0006456919360404924,1155 "loss": 0.117,1156 "step": 126721157 },1158 {1159 "epoch": 1.08,1160 "eval_loss": 0.08993188291788101,1161 "eval_max_distance": 82,1162 "eval_mean_distance": 2,1163 "eval_runtime": 8.194,1164 "eval_samples_per_second": 61.02,1165 "eval_steps_per_second": 1.22,1166 "step": 126721167 },1168 {1169 "epoch": 1.1,1170 "learning_rate": 0.0006406303922696422,1171 "loss": 0.1077,1172 "step": 128481173 },1174 {1175 "epoch": 1.1,1176 "eval_loss": 0.09431099146604538,1177 "eval_max_distance": 76,1178 "eval_mean_distance": 2,1179 "eval_runtime": 7.8825,1180 "eval_samples_per_second": 63.432,1181 "eval_steps_per_second": 1.269,1182 "step": 128481183 },1184 {1185 "epoch": 1.11,1186 "learning_rate": 0.0006355688484987922,1187 "loss": 0.1023,1188 "step": 130241189 },1190 {1191 "epoch": 1.11,1192 "eval_loss": 0.08808578550815582,1193 "eval_max_distance": 71,1194 "eval_mean_distance": 2,1195 "eval_runtime": 9.0408,1196 "eval_samples_per_second": 55.305,1197 "eval_steps_per_second": 1.106,1198 "step": 130241199 },1200 {