fals3/Salesforce-codegen2-1B_P-unit-test-lora
04
1{2 "best_metric": null,3 "best_model_checkpoint": null,4 "epoch": 2.995008319467554,5 "eval_steps": 500,6 "global_step": 1350,7 "is_hyper_param_search": false,8 "is_local_process_zero": true,9 "is_world_process_zero": true,10 "log_history": [11 {12 "epoch": 0.0022185246810870773,13 "grad_norm": 2.1247129440307617,14 "learning_rate": 2.222222222222222e-06,15 "loss": 9.699,16 "step": 117 },18 {19 "epoch": 0.004437049362174155,20 "grad_norm": 2.200881004333496,21 "learning_rate": 4.444444444444444e-06,22 "loss": 9.5438,23 "step": 224 },25 {26 "epoch": 0.0066555740432612314,27 "grad_norm": 2.15293288230896,28 "learning_rate": 6.666666666666667e-06,29 "loss": 9.5488,30 "step": 331 },32 {33 "epoch": 0.00887409872434831,34 "grad_norm": 2.195706844329834,35 "learning_rate": 8.888888888888888e-06,36 "loss": 9.4276,37 "step": 438 },39 {40 "epoch": 0.011092623405435386,41 "grad_norm": 2.062354803085327,42 "learning_rate": 1.111111111111111e-05,43 "loss": 9.4862,44 "step": 545 },46 {47 "epoch": 0.013311148086522463,48 "grad_norm": 2.240753650665283,49 "learning_rate": 1.3333333333333333e-05,50 "loss": 9.4234,51 "step": 652 },53 {54 "epoch": 0.01552967276760954,55 "grad_norm": 2.088193416595459,56 "learning_rate": 1.5555555555555555e-05,57 "loss": 9.4802,58 "step": 759 },60 {61 "epoch": 0.01774819744869662,62 "grad_norm": 2.2427570819854736,63 "learning_rate": 1.7777777777777777e-05,64 "loss": 9.7638,65 "step": 866 },67 {68 "epoch": 0.019966722129783693,69 "grad_norm": 2.54486346244812,70 "learning_rate": 1.9999999999999998e-05,71 "loss": 9.2772,72 "step": 973 },74 {75 "epoch": 0.022185246810870772,76 "grad_norm": 2.3003125190734863,77 "learning_rate": 2.222222222222222e-05,78 "loss": 9.6489,79 "step": 1080 },81 {82 "epoch": 0.024403771491957847,83 "grad_norm": 2.762918710708618,84 "learning_rate": 2.4444444444444445e-05,85 "loss": 9.3257,86 "step": 1187 },88 {89 "epoch": 0.026622296173044926,90 "grad_norm": 2.6459238529205322,91 "learning_rate": 2.6666666666666667e-05,92 "loss": 9.3666,93 "step": 1294 },95 {96 "epoch": 0.028840820854132,97 "grad_norm": 2.8060708045959473,98 "learning_rate": 2.8888888888888885e-05,99 "loss": 9.4591,100 "step": 13101 },102 {103 "epoch": 0.03105934553521908,104 "grad_norm": 2.7448670864105225,105 "learning_rate": 3.111111111111111e-05,106 "loss": 9.3005,107 "step": 14108 },109 {110 "epoch": 0.033277870216306155,111 "grad_norm": 2.9884493350982666,112 "learning_rate": 3.333333333333333e-05,113 "loss": 9.413,114 "step": 15115 },116 {117 "epoch": 0.03549639489739324,118 "grad_norm": 2.796663522720337,119 "learning_rate": 3.555555555555555e-05,120 "loss": 9.3922,121 "step": 16122 },123 {124 "epoch": 0.03771491957848031,125 "grad_norm": 3.1795897483825684,126 "learning_rate": 3.777777777777777e-05,127 "loss": 9.111,128 "step": 17129 },130 {131 "epoch": 0.03993344425956739,132 "grad_norm": 3.041274070739746,133 "learning_rate": 3.9999999999999996e-05,134 "loss": 9.0777,135 "step": 18136 },137 {138 "epoch": 0.04215196894065446,139 "grad_norm": 3.1814870834350586,140 "learning_rate": 4.222222222222222e-05,141 "loss": 9.0334,142 "step": 19143 },144 {145 "epoch": 0.044370493621741544,146 "grad_norm": 3.389146566390991,147 "learning_rate": 4.444444444444444e-05,148 "loss": 8.8284,149 "step": 20150 },151 {152 "epoch": 0.04658901830282862,153 "grad_norm": 3.4111945629119873,154 "learning_rate": 4.6666666666666665e-05,155 "loss": 8.6863,156 "step": 21157 },158 {159 "epoch": 0.048807542983915694,160 "grad_norm": 3.2762200832366943,161 "learning_rate": 4.888888888888889e-05,162 "loss": 8.4764,163 "step": 22164 },165 {166 "epoch": 0.051026067665002776,167 "grad_norm": 3.4036614894866943,168 "learning_rate": 5.111111111111111e-05,169 "loss": 8.7689,170 "step": 23171 },172 {173 "epoch": 0.05324459234608985,174 "grad_norm": 3.7457964420318604,175 "learning_rate": 5.333333333333333e-05,176 "loss": 8.7579,177 "step": 24178 },179 {180 "epoch": 0.05546311702717693,181 "grad_norm": 3.8012592792510986,182 "learning_rate": 5.5555555555555545e-05,183 "loss": 8.1657,184 "step": 25185 },186 {187 "epoch": 0.057681641708264,188 "grad_norm": 3.8158586025238037,189 "learning_rate": 5.777777777777777e-05,190 "loss": 8.2045,191 "step": 26192 },193 {194 "epoch": 0.059900166389351084,195 "grad_norm": 5.272186279296875,196 "learning_rate": 5.9999999999999995e-05,197 "loss": 8.2431,198 "step": 27199 },200 {201 "epoch": 0.06211869107043816,202 "grad_norm": 5.857285022735596,203 "learning_rate": 6.222222222222222e-05,204 "loss": 8.2881,205 "step": 28206 },207 {208 "epoch": 0.06433721575152523,209 "grad_norm": 5.007485866546631,210 "learning_rate": 6.444444444444444e-05,211 "loss": 8.0562,212 "step": 29213 },214 {215 "epoch": 0.06655574043261231,216 "grad_norm": 5.025998592376709,217 "learning_rate": 6.666666666666666e-05,218 "loss": 7.8378,219 "step": 30220 },221 {222 "epoch": 0.06877426511369938,223 "grad_norm": 4.809839248657227,224 "learning_rate": 6.888888888888888e-05,225 "loss": 7.7989,226 "step": 31227 },228 {229 "epoch": 0.07099278979478647,230 "grad_norm": 3.9237091541290283,231 "learning_rate": 7.11111111111111e-05,232 "loss": 7.5362,233 "step": 32234 },235 {236 "epoch": 0.07321131447587355,237 "grad_norm": 3.669914960861206,238 "learning_rate": 7.333333333333332e-05,239 "loss": 7.5813,240 "step": 33241 },242 {243 "epoch": 0.07542983915696062,244 "grad_norm": 3.283878803253174,245 "learning_rate": 7.555555555555554e-05,246 "loss": 7.3411,247 "step": 34248 },249 {250 "epoch": 0.0776483638380477,251 "grad_norm": 3.651585817337036,252 "learning_rate": 7.777777777777777e-05,253 "loss": 7.0364,254 "step": 35255 },256 {257 "epoch": 0.07986688851913477,258 "grad_norm": 3.2250444889068604,259 "learning_rate": 7.999999999999999e-05,260 "loss": 7.3274,261 "step": 36262 },263 {264 "epoch": 0.08208541320022185,265 "grad_norm": 3.6796436309814453,266 "learning_rate": 8.222222222222222e-05,267 "loss": 7.4356,268 "step": 37269 },270 {271 "epoch": 0.08430393788130892,272 "grad_norm": 4.444842338562012,273 "learning_rate": 8.444444444444444e-05,274 "loss": 7.0472,275 "step": 38276 },277 {278 "epoch": 0.08652246256239601,279 "grad_norm": 4.753952503204346,280 "learning_rate": 8.666666666666665e-05,281 "loss": 6.7329,282 "step": 39283 },284 {285 "epoch": 0.08874098724348309,286 "grad_norm": 3.6671204566955566,287 "learning_rate": 8.888888888888888e-05,288 "loss": 6.8531,289 "step": 40290 },291 {292 "epoch": 0.09095951192457016,293 "grad_norm": 2.9891927242279053,294 "learning_rate": 9.11111111111111e-05,295 "loss": 7.499,296 "step": 41297 },298 {299 "epoch": 0.09317803660565724,300 "grad_norm": 2.8181629180908203,301 "learning_rate": 9.333333333333333e-05,302 "loss": 6.6919,303 "step": 42304 },305 {306 "epoch": 0.09539656128674431,307 "grad_norm": 2.4321961402893066,308 "learning_rate": 9.555555555555555e-05,309 "loss": 6.7912,310 "step": 43311 },312 {313 "epoch": 0.09761508596783139,314 "grad_norm": 2.447136640548706,315 "learning_rate": 9.777777777777778e-05,316 "loss": 6.7424,317 "step": 44318 },319 {320 "epoch": 0.09983361064891846,321 "grad_norm": 2.347606897354126,322 "learning_rate": 9.999999999999999e-05,323 "loss": 6.5965,324 "step": 45325 },326 {327 "epoch": 0.10205213533000555,328 "grad_norm": 3.112799644470215,329 "learning_rate": 0.00010222222222222222,330 "loss": 6.9193,331 "step": 46332 },333 {334 "epoch": 0.10427066001109263,335 "grad_norm": 2.9997987747192383,336 "learning_rate": 0.00010444444444444444,337 "loss": 6.6834,338 "step": 47339 },340 {341 "epoch": 0.1064891846921797,342 "grad_norm": 2.2601795196533203,343 "learning_rate": 0.00010666666666666667,344 "loss": 6.4838,345 "step": 48346 },347 {348 "epoch": 0.10870770937326678,349 "grad_norm": 2.697308301925659,350 "learning_rate": 0.00010888888888888889,351 "loss": 6.7889,352 "step": 49353 },354 {355 "epoch": 0.11092623405435385,356 "grad_norm": 1.8948653936386108,357 "learning_rate": 0.00011111111111111109,358 "loss": 6.5378,359 "step": 50360 },361 {362 "epoch": 0.11314475873544093,363 "grad_norm": 2.4612224102020264,364 "learning_rate": 0.00011333333333333331,365 "loss": 6.4858,366 "step": 51367 },368 {369 "epoch": 0.115363283416528,370 "grad_norm": 1.8843343257904053,371 "learning_rate": 0.00011555555555555554,372 "loss": 6.4761,373 "step": 52374 },375 {376 "epoch": 0.11758180809761509,377 "grad_norm": 1.881941795349121,378 "learning_rate": 0.00011777777777777776,379 "loss": 6.6741,380 "step": 53381 },382 {383 "epoch": 0.11980033277870217,384 "grad_norm": 1.630995750427246,385 "learning_rate": 0.00011999999999999999,386 "loss": 6.5392,387 "step": 54388 },389 {390 "epoch": 0.12201885745978924,391 "grad_norm": 1.6674602031707764,392 "learning_rate": 0.00012222222222222221,393 "loss": 6.3921,394 "step": 55395 },396 {397 "epoch": 0.12423738214087632,398 "grad_norm": 1.7581100463867188,399 "learning_rate": 0.00012444444444444444,400 "loss": 6.3696,401 "step": 56402 },403 {404 "epoch": 0.1264559068219634,405 "grad_norm": 2.0795226097106934,406 "learning_rate": 0.00012666666666666666,407 "loss": 6.4294,408 "step": 57409 },410 {411 "epoch": 0.12867443150305047,412 "grad_norm": 1.7071822881698608,413 "learning_rate": 0.0001288888888888889,414 "loss": 6.3776,415 "step": 58416 },417 {418 "epoch": 0.13089295618413754,419 "grad_norm": 2.0405585765838623,420 "learning_rate": 0.00013111111111111111,421 "loss": 6.2884,422 "step": 59423 },424 {425 "epoch": 0.13311148086522462,426 "grad_norm": 1.8677833080291748,427 "learning_rate": 0.0001333333333333333,428 "loss": 6.1985,429 "step": 60430 },431 {432 "epoch": 0.1353300055463117,433 "grad_norm": 1.7754088640213013,434 "learning_rate": 0.00013555555555555554,435 "loss": 6.4073,436 "step": 61437 },438 {439 "epoch": 0.13754853022739877,440 "grad_norm": 1.7457290887832642,441 "learning_rate": 0.00013777777777777776,442 "loss": 6.4382,443 "step": 62444 },445 {446 "epoch": 0.13976705490848584,447 "grad_norm": 1.5886279344558716,448 "learning_rate": 0.00014,449 "loss": 6.4177,450 "step": 63451 },452 {453 "epoch": 0.14198557958957295,454 "grad_norm": 2.2759571075439453,455 "learning_rate": 0.0001422222222222222,456 "loss": 6.3328,457 "step": 64458 },459 {460 "epoch": 0.14420410427066002,461 "grad_norm": 1.9168124198913574,462 "learning_rate": 0.0001444444444444444,463 "loss": 6.3291,464 "step": 65465 },466 {467 "epoch": 0.1464226289517471,468 "grad_norm": 1.8469194173812866,469 "learning_rate": 0.00014666666666666664,470 "loss": 6.4009,471 "step": 66472 },473 {474 "epoch": 0.14864115363283417,475 "grad_norm": 1.7399849891662598,476 "learning_rate": 0.00014888888888888886,477 "loss": 6.3407,478 "step": 67479 },480 {481 "epoch": 0.15085967831392125,482 "grad_norm": 1.7461535930633545,483 "learning_rate": 0.00015111111111111109,484 "loss": 6.2734,485 "step": 68486 },487 {488 "epoch": 0.15307820299500832,489 "grad_norm": 2.3190548419952393,490 "learning_rate": 0.0001533333333333333,491 "loss": 6.1248,492 "step": 69493 },494 {495 "epoch": 0.1552967276760954,496 "grad_norm": 1.932100772857666,497 "learning_rate": 0.00015555555555555554,498 "loss": 6.1844,499 "step": 70500 },501 {502 "epoch": 0.15751525235718247,503 "grad_norm": 1.9892686605453491,504 "learning_rate": 0.00015777777777777776,505 "loss": 6.3337,506 "step": 71507 },508 {509 "epoch": 0.15973377703826955,510 "grad_norm": 1.996738076210022,511 "learning_rate": 0.00015999999999999999,512 "loss": 6.2516,513 "step": 72514 },515 {516 "epoch": 0.16195230171935662,517 "grad_norm": 1.7729567289352417,518 "learning_rate": 0.0001622222222222222,519 "loss": 6.2537,520 "step": 73521 },522 {523 "epoch": 0.1641708264004437,524 "grad_norm": 2.0228068828582764,525 "learning_rate": 0.00016444444444444444,526 "loss": 6.4105,527 "step": 74528 },529 {530 "epoch": 0.16638935108153077,531 "grad_norm": 2.1054394245147705,532 "learning_rate": 0.00016666666666666666,533 "loss": 6.0495,534 "step": 75535 },536 {537 "epoch": 0.16860787576261785,538 "grad_norm": 1.7938268184661865,539 "learning_rate": 0.00016888888888888889,540 "loss": 6.6331,541 "step": 76542 },543 {544 "epoch": 0.17082640044370492,545 "grad_norm": 1.801116943359375,546 "learning_rate": 0.0001711111111111111,547 "loss": 6.248,548 "step": 77549 },550 {551 "epoch": 0.17304492512479203,552 "grad_norm": 2.2520036697387695,553 "learning_rate": 0.0001733333333333333,554 "loss": 6.1265,555 "step": 78556 },557 {558 "epoch": 0.1752634498058791,559 "grad_norm": 2.5786898136138916,560 "learning_rate": 0.00017555555555555553,561 "loss": 6.4172,562 "step": 79563 },564 {565 "epoch": 0.17748197448696618,566 "grad_norm": 2.454434394836426,567 "learning_rate": 0.00017777777777777776,568 "loss": 6.2877,569 "step": 80570 },571 {572 "epoch": 0.17970049916805325,573 "grad_norm": 1.8599268198013306,574 "learning_rate": 0.00017999999999999998,575 "loss": 6.2774,576 "step": 81577 },578 {579 "epoch": 0.18191902384914033,580 "grad_norm": 3.30815052986145,581 "learning_rate": 0.0001822222222222222,582 "loss": 6.2997,583 "step": 82584 },585 {586 "epoch": 0.1841375485302274,587 "grad_norm": 2.632484197616577,588 "learning_rate": 0.00018444444444444443,589 "loss": 6.1476,590 "step": 83591 },592 {593 "epoch": 0.18635607321131448,594 "grad_norm": 2.256540060043335,595 "learning_rate": 0.00018666666666666666,596 "loss": 6.1887,597 "step": 84598 },599 {600 "epoch": 0.18857459789240155,601 "grad_norm": 2.5506718158721924,602 "learning_rate": 0.00018888888888888888,603 "loss": 6.1926,604 "step": 85605 },606 {607 "epoch": 0.19079312257348863,608 "grad_norm": 2.3097774982452393,609 "learning_rate": 0.0001911111111111111,610 "loss": 5.961,611 "step": 86612 },613 {614 "epoch": 0.1930116472545757,615 "grad_norm": 3.216681480407715,616 "learning_rate": 0.00019333333333333333,617 "loss": 6.2499,618 "step": 87619 },620 {621 "epoch": 0.19523017193566278,622 "grad_norm": 2.977572202682495,623 "learning_rate": 0.00019555555555555556,624 "loss": 6.0447,625 "step": 88626 },627 {628 "epoch": 0.19744869661674985,629 "grad_norm": 4.106575965881348,630 "learning_rate": 0.00019777777777777776,631 "loss": 6.3153,632 "step": 89633 },634 {635 "epoch": 0.19966722129783693,636 "grad_norm": 2.4921114444732666,637 "learning_rate": 0.00019999999999999998,638 "loss": 6.1619,639 "step": 90640 },641 {642 "epoch": 0.201885745978924,643 "grad_norm": 2.3812241554260254,644 "learning_rate": 0.0002022222222222222,645 "loss": 6.1412,646 "step": 91647 },648 {649 "epoch": 0.2041042706600111,650 "grad_norm": 3.820882797241211,651 "learning_rate": 0.00020444444444444443,652 "loss": 6.0223,653 "step": 92654 },655 {656 "epoch": 0.20632279534109818,657 "grad_norm": 3.3213260173797607,658 "learning_rate": 0.00020666666666666666,659 "loss": 6.2297,660 "step": 93661 },662 {663 "epoch": 0.20854132002218526,664 "grad_norm": 3.2521910667419434,665 "learning_rate": 0.00020888888888888888,666 "loss": 6.2223,667 "step": 94668 },669 {670 "epoch": 0.21075984470327233,671 "grad_norm": 3.3431460857391357,672 "learning_rate": 0.0002111111111111111,673 "loss": 6.0618,674 "step": 95675 },676 {677 "epoch": 0.2129783693843594,678 "grad_norm": 3.6344332695007324,679 "learning_rate": 0.00021333333333333333,680 "loss": 6.1421,681 "step": 96682 },683 {684 "epoch": 0.21519689406544648,685 "grad_norm": 2.366407871246338,686 "learning_rate": 0.00021555555555555556,687 "loss": 5.8652,688 "step": 97689 },690 {691 "epoch": 0.21741541874653356,692 "grad_norm": 5.667245388031006,693 "learning_rate": 0.00021777777777777778,694 "loss": 5.9991,695 "step": 98696 },697 {698 "epoch": 0.21963394342762063,699 "grad_norm": 3.6710238456726074,700 "learning_rate": 0.00021999999999999995,701 "loss": 6.1888,702 "step": 99703 },704 {705 "epoch": 0.2218524681087077,706 "grad_norm": 7.917342185974121,707 "learning_rate": 0.00022222222222222218,708 "loss": 5.8296,709 "step": 100710 },711 {712 "epoch": 0.22407099278979478,713 "grad_norm": 7.94102144241333,714 "learning_rate": 0.0002244444444444444,715 "loss": 5.4338,716 "step": 101717 },718 {719 "epoch": 0.22628951747088186,720 "grad_norm": 2.975712299346924,721 "learning_rate": 0.00022666666666666663,722 "loss": 5.7384,723 "step": 102724 },725 {726 "epoch": 0.22850804215196893,727 "grad_norm": 10.478729248046875,728 "learning_rate": 0.00022888888888888885,729 "loss": 5.9641,730 "step": 103731 },732 {733 "epoch": 0.230726566833056,734 "grad_norm": 11.62560749053955,735 "learning_rate": 0.00023111111111111108,736 "loss": 5.4342,737 "step": 104738 },739 {740 "epoch": 0.23294509151414308,741 "grad_norm": 5.349384784698486,742 "learning_rate": 0.0002333333333333333,743 "loss": 5.8547,744 "step": 105745 },746 {747 "epoch": 0.23516361619523019,748 "grad_norm": 8.264240264892578,749 "learning_rate": 0.00023555555555555553,750 "loss": 5.7199,751 "step": 106752 },753 {754 "epoch": 0.23738214087631726,755 "grad_norm": 12.192222595214844,756 "learning_rate": 0.00023777777777777775,757 "loss": 5.6058,758 "step": 107759 },760 {761 "epoch": 0.23960066555740434,762 "grad_norm": 4.782373428344727,763 "learning_rate": 0.00023999999999999998,764 "loss": 5.7377,765 "step": 108766 },767 {768 "epoch": 0.2418191902384914,769 "grad_norm": 9.63099193572998,770 "learning_rate": 0.0002422222222222222,771 "loss": 5.7224,772 "step": 109773 },774 {775 "epoch": 0.24403771491957849,776 "grad_norm": 11.105454444885254,777 "learning_rate": 0.00024444444444444443,778 "loss": 5.8202,779 "step": 110780 },781 {782 "epoch": 0.24625623960066556,783 "grad_norm": 4.963940143585205,784 "learning_rate": 0.0002466666666666666,785 "loss": 5.7028,786 "step": 111787 },788 {789 "epoch": 0.24847476428175264,790 "grad_norm": 9.338000297546387,791 "learning_rate": 0.0002488888888888889,792 "loss": 5.4165,793 "step": 112794 },795 {796 "epoch": 0.2506932889628397,797 "grad_norm": 9.696998596191406,798 "learning_rate": 0.0002511111111111111,799 "loss": 5.4409,800 "step": 113801 },802 {803 "epoch": 0.2529118136439268,804 "grad_norm": 5.044449329376221,805 "learning_rate": 0.00025333333333333333,806 "loss": 5.4261,807 "step": 114808 },809 {810 "epoch": 0.25513033832501386,811 "grad_norm": 5.547409534454346,812 "learning_rate": 0.00025555555555555553,813 "loss": 5.3918,814 "step": 115815 },816 {817 "epoch": 0.25734886300610094,818 "grad_norm": 5.158348083496094,819 "learning_rate": 0.0002577777777777778,820 "loss": 5.4578,821 "step": 116822 },823 {824 "epoch": 0.259567387687188,825 "grad_norm": 4.160797595977783,826 "learning_rate": 0.00026,827 "loss": 5.3358,828 "step": 117829 },830 {831 "epoch": 0.2617859123682751,832 "grad_norm": 7.16011381149292,833 "learning_rate": 0.00026222222222222223,834 "loss": 5.3624,835 "step": 118836 },837 {838 "epoch": 0.26400443704936216,839 "grad_norm": 5.639223575592041,840 "learning_rate": 0.00026444444444444443,841 "loss": 5.248,842 "step": 119843 },844 {845 "epoch": 0.26622296173044924,846 "grad_norm": 8.2030029296875,847 "learning_rate": 0.0002666666666666666,848 "loss": 5.3996,849 "step": 120850 },851 {852 "epoch": 0.2684414864115363,853 "grad_norm": 6.017608165740967,854 "learning_rate": 0.0002688888888888889,855 "loss": 5.0931,856 "step": 121857 },858 {859 "epoch": 0.2706600110926234,860 "grad_norm": 9.665661811828613,861 "learning_rate": 0.0002711111111111111,862 "loss": 5.3154,863 "step": 122864 },865 {866 "epoch": 0.27287853577371046,867 "grad_norm": 8.194005012512207,868 "learning_rate": 0.00027333333333333333,869 "loss": 5.1912,870 "step": 123871 },872 {873 "epoch": 0.27509706045479754,874 "grad_norm": 7.089859485626221,875 "learning_rate": 0.0002755555555555555,876 "loss": 5.2339,877 "step": 124878 },879 {880 "epoch": 0.2773155851358846,881 "grad_norm": 11.654187202453613,882 "learning_rate": 0.0002777777777777778,883 "loss": 5.4888,884 "step": 125885 },886 {887 "epoch": 0.2795341098169717,888 "grad_norm": 9.684091567993164,889 "learning_rate": 0.00028,890 "loss": 5.282,891 "step": 126892 },893 {894 "epoch": 0.2817526344980588,895 "grad_norm": 10.477103233337402,896 "learning_rate": 0.00028222222222222223,897 "loss": 5.205,898 "step": 127899 },900 {901 "epoch": 0.2839711591791459,902 "grad_norm": 8.727799415588379,903 "learning_rate": 0.0002844444444444444,904 "loss": 5.4352,905 "step": 128906 },907 {908 "epoch": 0.28618968386023297,909 "grad_norm": 8.56086254119873,910 "learning_rate": 0.0002866666666666667,911 "loss": 5.2909,912 "step": 129913 },914 {915 "epoch": 0.28840820854132004,916 "grad_norm": 8.4285249710083,917 "learning_rate": 0.0002888888888888888,918 "loss": 5.3951,919 "step": 130920 },921 {922 "epoch": 0.2906267332224071,923 "grad_norm": 6.3976054191589355,924 "learning_rate": 0.0002911111111111111,925 "loss": 5.4093,926 "step": 131927 },928 {929 "epoch": 0.2928452579034942,930 "grad_norm": 6.13315486907959,931 "learning_rate": 0.00029333333333333327,932 "loss": 4.922,933 "step": 132934 },935 {936 "epoch": 0.29506378258458127,937 "grad_norm": 6.330722808837891,938 "learning_rate": 0.0002955555555555555,939 "loss": 5.2876,940 "step": 133941 },942 {943 "epoch": 0.29728230726566834,944 "grad_norm": 6.6806135177612305,945 "learning_rate": 0.0002977777777777777,946 "loss": 5.051,947 "step": 134948 },949 {950 "epoch": 0.2995008319467554,951 "grad_norm": 5.428084373474121,952 "learning_rate": 0.0003,953 "loss": 4.9329,954 "step": 135955 },956 {957 "epoch": 0.3017193566278425,958 "grad_norm": 6.041404724121094,959 "learning_rate": 0.00029975308641975307,960 "loss": 4.8975,961 "step": 136962 },963 {964 "epoch": 0.30393788130892957,965 "grad_norm": 4.520870685577393,966 "learning_rate": 0.00029950617283950616,967 "loss": 5.3007,968 "step": 137969 },970 {971 "epoch": 0.30615640599001664,972 "grad_norm": 9.24893856048584,973 "learning_rate": 0.00029925925925925926,974 "loss": 5.134,975 "step": 138976 },977 {978 "epoch": 0.3083749306711037,979 "grad_norm": 6.243694305419922,980 "learning_rate": 0.0002990123456790123,981 "loss": 5.0237,982 "step": 139983 },984 {985 "epoch": 0.3105934553521908,986 "grad_norm": 4.546097278594971,987 "learning_rate": 0.0002987654320987654,988 "loss": 5.0313,989 "step": 140990 },991 {992 "epoch": 0.31281198003327787,993 "grad_norm": 5.99699592590332,994 "learning_rate": 0.0002985185185185185,995 "loss": 4.8896,996 "step": 141997 },998 {999 "epoch": 0.31503050471436495,1000 "grad_norm": 8.150283813476562,1001 "learning_rate": 0.0002982716049382716,1002 "loss": 5.2722,1003 "step": 1421004 },1005 {1006 "epoch": 0.317249029395452,1007 "grad_norm": 7.2810258865356445,1008 "learning_rate": 0.0002980246913580247,1009 "loss": 4.9425,1010 "step": 1431011 },1012 {1013 "epoch": 0.3194675540765391,1014 "grad_norm": 4.84526252746582,1015 "learning_rate": 0.0002977777777777777,1016 "loss": 5.1888,1017 "step": 1441018 },1019 {1020 "epoch": 0.32168607875762617,1021 "grad_norm": 6.3347954750061035,1022 "learning_rate": 0.00029753086419753087,1023 "loss": 5.0802,1024 "step": 1451025 },1026 {1027 "epoch": 0.32390460343871325,1028 "grad_norm": 5.987547397613525,1029 "learning_rate": 0.0002972839506172839,1030 "loss": 4.9879,1031 "step": 1461032 },1033 {1034 "epoch": 0.3261231281198003,1035 "grad_norm": 5.489535331726074,1036 "learning_rate": 0.000297037037037037,1037 "loss": 4.7362,1038 "step": 1471039 },1040 {1041 "epoch": 0.3283416528008874,1042 "grad_norm": 9.252480506896973,1043 "learning_rate": 0.0002967901234567901,1044 "loss": 5.2288,1045 "step": 1481046 },1047 {1048 "epoch": 0.33056017748197447,1049 "grad_norm": 8.22285270690918,1050 "learning_rate": 0.0002965432098765432,1051 "loss": 5.0849,1052 "step": 1491053 },1054 {1055 "epoch": 0.33277870216306155,1056 "grad_norm": 5.304591655731201,1057 "learning_rate": 0.00029629629629629624,1058 "loss": 4.9756,1059 "step": 1501060 },1061 {1062 "epoch": 0.3349972268441486,1063 "grad_norm": 7.345332622528076,1064 "learning_rate": 0.00029604938271604933,1065 "loss": 5.1705,1066 "step": 1511067 },1068 {1069 "epoch": 0.3372157515252357,1070 "grad_norm": 4.644253253936768,1071 "learning_rate": 0.0002958024691358025,1072 "loss": 5.0919,1073 "step": 1521074 },1075 {1076 "epoch": 0.33943427620632277,1077 "grad_norm": 3.943446636199951,1078 "learning_rate": 0.0002955555555555555,1079 "loss": 4.9817,1080 "step": 1531081 },1082 {1083 "epoch": 0.34165280088740985,1084 "grad_norm": 4.669249057769775,1085 "learning_rate": 0.0002953086419753086,1086 "loss": 4.9341,1087 "step": 1541088 },1089 {1090 "epoch": 0.343871325568497,1091 "grad_norm": 5.717892646789551,1092 "learning_rate": 0.0002950617283950617,1093 "loss": 5.1695,1094 "step": 1551095 },1096 {1097 "epoch": 0.34608985024958405,1098 "grad_norm": 4.880568027496338,1099 "learning_rate": 0.0002948148148148148,1100 "loss": 4.9563,1101 "step": 1561102 },1103 {1104 "epoch": 0.3483083749306711,1105 "grad_norm": 4.0522990226745605,1106 "learning_rate": 0.00029456790123456785,1107 "loss": 5.0349,1108 "step": 1571109 },1110 {1111 "epoch": 0.3505268996117582,1112 "grad_norm": 5.536722660064697,1113 "learning_rate": 0.00029432098765432095,1114 "loss": 5.1229,1115 "step": 1581116 },1117 {1118 "epoch": 0.3527454242928453,1119 "grad_norm": 3.4449219703674316,1120 "learning_rate": 0.00029407407407407404,1121 "loss": 4.5567,1122 "step": 1591123 },1124 {1125 "epoch": 0.35496394897393235,1126 "grad_norm": 4.434334754943848,1127 "learning_rate": 0.00029382716049382714,1128 "loss": 4.7849,1129 "step": 1601130 },1131 {1132 "epoch": 0.35718247365501943,1133 "grad_norm": 5.1075544357299805,1134 "learning_rate": 0.00029358024691358023,1135 "loss": 5.3021,1136 "step": 1611137 },1138 {1139 "epoch": 0.3594009983361065,1140 "grad_norm": 5.96935510635376,1141 "learning_rate": 0.00029333333333333327,1142 "loss": 5.0779,1143 "step": 1621144 },1145 {1146 "epoch": 0.3616195230171936,1147 "grad_norm": 6.567812919616699,1148 "learning_rate": 0.0002930864197530864,1149 "loss": 5.0314,1150 "step": 1631151 },1152 {1153 "epoch": 0.36383804769828065,1154 "grad_norm": 5.427769184112549,1155 "learning_rate": 0.00029283950617283946,1156 "loss": 4.9587,1157 "step": 1641158 },1159 {1160 "epoch": 0.36605657237936773,1161 "grad_norm": 7.360152721405029,1162 "learning_rate": 0.00029259259259259256,1163 "loss": 4.966,1164 "step": 1651165 },1166 {1167 "epoch": 0.3682750970604548,1168 "grad_norm": 5.89814567565918,1169 "learning_rate": 0.00029234567901234565,1170 "loss": 5.0995,1171 "step": 1661172 },1173 {1174 "epoch": 0.3704936217415419,1175 "grad_norm": 5.960623741149902,1176 "learning_rate": 0.00029209876543209875,1177 "loss": 4.7687,1178 "step": 1671179 },1180 {1181 "epoch": 0.37271214642262895,1182 "grad_norm": 4.919917583465576,1183 "learning_rate": 0.00029185185185185184,1184 "loss": 4.8205,1185 "step": 1681186 },1187 {1188 "epoch": 0.37493067110371603,1189 "grad_norm": 6.094188690185547,1190 "learning_rate": 0.0002916049382716049,1191 "loss": 4.667,1192 "step": 1691193 },1194 {1195 "epoch": 0.3771491957848031,1196 "grad_norm": 5.920214653015137,1197 "learning_rate": 0.000291358024691358,1198 "loss": 4.7728,1199 "step": 1701200 },