dipikakhullar/olmo-code-python3-chunk-aa
09
1{2 "best_global_step": null,3 "best_metric": null,4 "best_model_checkpoint": null,5 "epoch": 1.0,6 "eval_steps": 500,7 "global_step": 11250,8 "is_hyper_param_search": false,9 "is_local_process_zero": true,10 "is_world_process_zero": true,11 "log_history": [12 {13 "epoch": 8.888888888888889e-05,14 "grad_norm": 2.3531012535095215,15 "learning_rate": 5e-05,16 "loss": 1.5211,17 "step": 118 },19 {20 "epoch": 0.00017777777777777779,21 "grad_norm": 3.995047092437744,22 "learning_rate": 4.999555555555556e-05,23 "loss": 1.5564,24 "step": 225 },26 {27 "epoch": 0.0002666666666666667,28 "grad_norm": 3.648165225982666,29 "learning_rate": 4.999111111111111e-05,30 "loss": 1.7195,31 "step": 332 },33 {34 "epoch": 0.00035555555555555557,35 "grad_norm": 4.093322277069092,36 "learning_rate": 4.9986666666666674e-05,37 "loss": 1.8164,38 "step": 439 },40 {41 "epoch": 0.00044444444444444447,42 "grad_norm": 2.33286190032959,43 "learning_rate": 4.998222222222222e-05,44 "loss": 1.7287,45 "step": 546 },47 {48 "epoch": 0.0005333333333333334,49 "grad_norm": 2.6810221672058105,50 "learning_rate": 4.997777777777778e-05,51 "loss": 1.8443,52 "step": 653 },54 {55 "epoch": 0.0006222222222222223,56 "grad_norm": 2.499145030975342,57 "learning_rate": 4.997333333333333e-05,58 "loss": 1.8891,59 "step": 760 },61 {62 "epoch": 0.0007111111111111111,63 "grad_norm": 2.9107909202575684,64 "learning_rate": 4.996888888888889e-05,65 "loss": 1.8394,66 "step": 867 },68 {69 "epoch": 0.0008,70 "grad_norm": 2.6283602714538574,71 "learning_rate": 4.996444444444445e-05,72 "loss": 1.8876,73 "step": 974 },75 {76 "epoch": 0.0008888888888888889,77 "grad_norm": 2.464566469192505,78 "learning_rate": 4.996e-05,79 "loss": 1.9369,80 "step": 1081 },82 {83 "epoch": 0.0009777777777777777,84 "grad_norm": 2.311979293823242,85 "learning_rate": 4.995555555555556e-05,86 "loss": 1.6825,87 "step": 1188 },89 {90 "epoch": 0.0010666666666666667,91 "grad_norm": 4.183473110198975,92 "learning_rate": 4.995111111111111e-05,93 "loss": 2.0672,94 "step": 1295 },96 {97 "epoch": 0.0011555555555555555,98 "grad_norm": 2.7625925540924072,99 "learning_rate": 4.994666666666667e-05,100 "loss": 1.6858,101 "step": 13102 },103 {104 "epoch": 0.0012444444444444445,105 "grad_norm": 2.9213318824768066,106 "learning_rate": 4.994222222222222e-05,107 "loss": 1.8734,108 "step": 14109 },110 {111 "epoch": 0.0013333333333333333,112 "grad_norm": 2.5807080268859863,113 "learning_rate": 4.993777777777778e-05,114 "loss": 1.9485,115 "step": 15116 },117 {118 "epoch": 0.0014222222222222223,119 "grad_norm": 2.355184316635132,120 "learning_rate": 4.993333333333334e-05,121 "loss": 1.7631,122 "step": 16123 },124 {125 "epoch": 0.001511111111111111,126 "grad_norm": 2.2967448234558105,127 "learning_rate": 4.9928888888888893e-05,128 "loss": 1.9014,129 "step": 17130 },131 {132 "epoch": 0.0016,133 "grad_norm": 2.1532392501831055,134 "learning_rate": 4.992444444444445e-05,135 "loss": 1.845,136 "step": 18137 },138 {139 "epoch": 0.0016888888888888889,140 "grad_norm": 2.44671630859375,141 "learning_rate": 4.992e-05,142 "loss": 1.8097,143 "step": 19144 },145 {146 "epoch": 0.0017777777777777779,147 "grad_norm": 2.6243488788604736,148 "learning_rate": 4.991555555555556e-05,149 "loss": 1.9376,150 "step": 20151 },152 {153 "epoch": 0.0018666666666666666,154 "grad_norm": 2.451446533203125,155 "learning_rate": 4.991111111111111e-05,156 "loss": 1.9679,157 "step": 21158 },159 {160 "epoch": 0.0019555555555555554,161 "grad_norm": 2.190685510635376,162 "learning_rate": 4.990666666666667e-05,163 "loss": 1.7698,164 "step": 22165 },166 {167 "epoch": 0.0020444444444444447,168 "grad_norm": 2.1394426822662354,169 "learning_rate": 4.990222222222222e-05,170 "loss": 1.7086,171 "step": 23172 },173 {174 "epoch": 0.0021333333333333334,175 "grad_norm": 2.048832893371582,176 "learning_rate": 4.9897777777777784e-05,177 "loss": 1.7068,178 "step": 24179 },180 {181 "epoch": 0.0022222222222222222,182 "grad_norm": 2.3317484855651855,183 "learning_rate": 4.989333333333334e-05,184 "loss": 1.8647,185 "step": 25186 },187 {188 "epoch": 0.002311111111111111,189 "grad_norm": 2.1953907012939453,190 "learning_rate": 4.9888888888888894e-05,191 "loss": 1.9136,192 "step": 26193 },194 {195 "epoch": 0.0024,196 "grad_norm": 2.451587677001953,197 "learning_rate": 4.988444444444444e-05,198 "loss": 1.8417,199 "step": 27200 },201 {202 "epoch": 0.002488888888888889,203 "grad_norm": 2.759969472885132,204 "learning_rate": 4.9880000000000004e-05,205 "loss": 1.8546,206 "step": 28207 },208 {209 "epoch": 0.002577777777777778,210 "grad_norm": 2.5941691398620605,211 "learning_rate": 4.987555555555556e-05,212 "loss": 1.912,213 "step": 29214 },215 {216 "epoch": 0.0026666666666666666,217 "grad_norm": 2.656020402908325,218 "learning_rate": 4.987111111111111e-05,219 "loss": 1.5926,220 "step": 30221 },222 {223 "epoch": 0.0027555555555555554,224 "grad_norm": 2.4222757816314697,225 "learning_rate": 4.986666666666667e-05,226 "loss": 1.8182,227 "step": 31228 },229 {230 "epoch": 0.0028444444444444446,231 "grad_norm": 2.9247982501983643,232 "learning_rate": 4.986222222222223e-05,233 "loss": 1.8425,234 "step": 32235 },236 {237 "epoch": 0.0029333333333333334,238 "grad_norm": 3.6813204288482666,239 "learning_rate": 4.985777777777778e-05,240 "loss": 1.6821,241 "step": 33242 },243 {244 "epoch": 0.003022222222222222,245 "grad_norm": 2.420562982559204,246 "learning_rate": 4.985333333333333e-05,247 "loss": 1.8912,248 "step": 34249 },250 {251 "epoch": 0.003111111111111111,252 "grad_norm": 2.2838096618652344,253 "learning_rate": 4.984888888888889e-05,254 "loss": 1.837,255 "step": 35256 },257 {258 "epoch": 0.0032,259 "grad_norm": 2.318554401397705,260 "learning_rate": 4.984444444444445e-05,261 "loss": 1.7914,262 "step": 36263 },264 {265 "epoch": 0.003288888888888889,266 "grad_norm": 2.3181023597717285,267 "learning_rate": 4.9840000000000004e-05,268 "loss": 1.694,269 "step": 37270 },271 {272 "epoch": 0.0033777777777777777,273 "grad_norm": 2.4494924545288086,274 "learning_rate": 4.983555555555556e-05,275 "loss": 1.9969,276 "step": 38277 },278 {279 "epoch": 0.0034666666666666665,280 "grad_norm": 2.1924023628234863,281 "learning_rate": 4.9831111111111114e-05,282 "loss": 1.8156,283 "step": 39284 },285 {286 "epoch": 0.0035555555555555557,287 "grad_norm": 2.3793368339538574,288 "learning_rate": 4.982666666666667e-05,289 "loss": 1.7035,290 "step": 40291 },292 {293 "epoch": 0.0036444444444444445,294 "grad_norm": 2.2038633823394775,295 "learning_rate": 4.982222222222222e-05,296 "loss": 1.7599,297 "step": 41298 },299 {300 "epoch": 0.0037333333333333333,301 "grad_norm": 1.9155387878417969,302 "learning_rate": 4.981777777777778e-05,303 "loss": 1.8145,304 "step": 42305 },306 {307 "epoch": 0.003822222222222222,308 "grad_norm": 1.9543706178665161,309 "learning_rate": 4.981333333333333e-05,310 "loss": 1.7996,311 "step": 43312 },313 {314 "epoch": 0.003911111111111111,315 "grad_norm": 3.6194260120391846,316 "learning_rate": 4.9808888888888895e-05,317 "loss": 1.9704,318 "step": 44319 },320 {321 "epoch": 0.004,322 "grad_norm": 2.3198981285095215,323 "learning_rate": 4.980444444444445e-05,324 "loss": 1.814,325 "step": 45326 },327 {328 "epoch": 0.004088888888888889,329 "grad_norm": 2.3419625759124756,330 "learning_rate": 4.9800000000000004e-05,331 "loss": 1.619,332 "step": 46333 },334 {335 "epoch": 0.004177777777777778,336 "grad_norm": 2.2805051803588867,337 "learning_rate": 4.979555555555556e-05,338 "loss": 1.7957,339 "step": 47340 },341 {342 "epoch": 0.004266666666666667,343 "grad_norm": 2.3934669494628906,344 "learning_rate": 4.9791111111111114e-05,345 "loss": 1.8718,346 "step": 48347 },348 {349 "epoch": 0.004355555555555555,350 "grad_norm": 2.138780117034912,351 "learning_rate": 4.978666666666667e-05,352 "loss": 1.8475,353 "step": 49354 },355 {356 "epoch": 0.0044444444444444444,357 "grad_norm": 2.322488784790039,358 "learning_rate": 4.9782222222222224e-05,359 "loss": 1.717,360 "step": 50361 },362 {363 "epoch": 0.004533333333333334,364 "grad_norm": 2.0249195098876953,365 "learning_rate": 4.977777777777778e-05,366 "loss": 1.9432,367 "step": 51368 },369 {370 "epoch": 0.004622222222222222,371 "grad_norm": 2.370011806488037,372 "learning_rate": 4.977333333333334e-05,373 "loss": 1.7384,374 "step": 52375 },376 {377 "epoch": 0.004711111111111111,378 "grad_norm": 2.1237905025482178,379 "learning_rate": 4.9768888888888895e-05,380 "loss": 1.5549,381 "step": 53382 },383 {384 "epoch": 0.0048,385 "grad_norm": 2.281357526779175,386 "learning_rate": 4.976444444444445e-05,387 "loss": 1.673,388 "step": 54389 },390 {391 "epoch": 0.004888888888888889,392 "grad_norm": 1.9526458978652954,393 "learning_rate": 4.976e-05,394 "loss": 1.7677,395 "step": 55396 },397 {398 "epoch": 0.004977777777777778,399 "grad_norm": 4.4257283210754395,400 "learning_rate": 4.975555555555555e-05,401 "loss": 2.2159,402 "step": 56403 },404 {405 "epoch": 0.005066666666666666,406 "grad_norm": 2.277620315551758,407 "learning_rate": 4.9751111111111114e-05,408 "loss": 1.5635,409 "step": 57410 },411 {412 "epoch": 0.005155555555555556,413 "grad_norm": 2.020883798599243,414 "learning_rate": 4.974666666666667e-05,415 "loss": 1.6738,416 "step": 58417 },418 {419 "epoch": 0.005244444444444445,420 "grad_norm": 5.044859886169434,421 "learning_rate": 4.9742222222222224e-05,422 "loss": 2.2399,423 "step": 59424 },425 {426 "epoch": 0.005333333333333333,427 "grad_norm": 2.468468189239502,428 "learning_rate": 4.973777777777778e-05,429 "loss": 1.772,430 "step": 60431 },432 {433 "epoch": 0.005422222222222222,434 "grad_norm": 1.9879132509231567,435 "learning_rate": 4.973333333333334e-05,436 "loss": 1.8662,437 "step": 61438 },439 {440 "epoch": 0.005511111111111111,441 "grad_norm": 2.234318494796753,442 "learning_rate": 4.972888888888889e-05,443 "loss": 1.8428,444 "step": 62445 },446 {447 "epoch": 0.0056,448 "grad_norm": 2.0144824981689453,449 "learning_rate": 4.9724444444444443e-05,450 "loss": 1.7328,451 "step": 63452 },453 {454 "epoch": 0.005688888888888889,455 "grad_norm": 3.1329941749572754,456 "learning_rate": 4.972e-05,457 "loss": 2.0239,458 "step": 64459 },460 {461 "epoch": 0.0057777777777777775,462 "grad_norm": 2.3503899574279785,463 "learning_rate": 4.971555555555556e-05,464 "loss": 1.8358,465 "step": 65466 },467 {468 "epoch": 0.005866666666666667,469 "grad_norm": 2.167900562286377,470 "learning_rate": 4.9711111111111115e-05,471 "loss": 1.8541,472 "step": 66473 },474 {475 "epoch": 0.005955555555555556,476 "grad_norm": 2.550446033477783,477 "learning_rate": 4.970666666666667e-05,478 "loss": 1.6915,479 "step": 67480 },481 {482 "epoch": 0.006044444444444444,483 "grad_norm": 2.2864279747009277,484 "learning_rate": 4.9702222222222224e-05,485 "loss": 1.78,486 "step": 68487 },488 {489 "epoch": 0.0061333333333333335,490 "grad_norm": 2.176593780517578,491 "learning_rate": 4.969777777777778e-05,492 "loss": 1.9499,493 "step": 69494 },495 {496 "epoch": 0.006222222222222222,497 "grad_norm": 2.358780860900879,498 "learning_rate": 4.9693333333333334e-05,499 "loss": 1.5643,500 "step": 70501 },502 {503 "epoch": 0.006311111111111111,504 "grad_norm": 1.9860084056854248,505 "learning_rate": 4.968888888888889e-05,506 "loss": 1.7757,507 "step": 71508 },509 {510 "epoch": 0.0064,511 "grad_norm": 2.1823506355285645,512 "learning_rate": 4.9684444444444444e-05,513 "loss": 1.7696,514 "step": 72515 },516 {517 "epoch": 0.006488888888888889,518 "grad_norm": 2.070911407470703,519 "learning_rate": 4.9680000000000005e-05,520 "loss": 1.7787,521 "step": 73522 },523 {524 "epoch": 0.006577777777777778,525 "grad_norm": 2.484304189682007,526 "learning_rate": 4.967555555555556e-05,527 "loss": 1.8739,528 "step": 74529 },530 {531 "epoch": 0.006666666666666667,532 "grad_norm": 2.85019588470459,533 "learning_rate": 4.9671111111111115e-05,534 "loss": 1.8429,535 "step": 75536 },537 {538 "epoch": 0.0067555555555555554,539 "grad_norm": 2.184279441833496,540 "learning_rate": 4.966666666666667e-05,541 "loss": 1.9036,542 "step": 76543 },544 {545 "epoch": 0.006844444444444445,546 "grad_norm": 2.5326406955718994,547 "learning_rate": 4.9662222222222225e-05,548 "loss": 1.8299,549 "step": 77550 },551 {552 "epoch": 0.006933333333333333,553 "grad_norm": 2.6205625534057617,554 "learning_rate": 4.965777777777778e-05,555 "loss": 1.8981,556 "step": 78557 },558 {559 "epoch": 0.007022222222222222,560 "grad_norm": 3.513176679611206,561 "learning_rate": 4.9653333333333335e-05,562 "loss": 1.8652,563 "step": 79564 },565 {566 "epoch": 0.0071111111111111115,567 "grad_norm": 2.5633668899536133,568 "learning_rate": 4.964888888888889e-05,569 "loss": 1.83,570 "step": 80571 },572 {573 "epoch": 0.0072,574 "grad_norm": 2.231003522872925,575 "learning_rate": 4.964444444444445e-05,576 "loss": 1.7978,577 "step": 81578 },579 {580 "epoch": 0.007288888888888889,581 "grad_norm": 2.1591978073120117,582 "learning_rate": 4.9640000000000006e-05,583 "loss": 1.6713,584 "step": 82585 },586 {587 "epoch": 0.007377777777777777,588 "grad_norm": 2.126413583755493,589 "learning_rate": 4.963555555555556e-05,590 "loss": 1.8979,591 "step": 83592 },593 {594 "epoch": 0.007466666666666667,595 "grad_norm": 2.6277430057525635,596 "learning_rate": 4.963111111111111e-05,597 "loss": 1.9564,598 "step": 84599 },600 {601 "epoch": 0.007555555555555556,602 "grad_norm": 2.076350212097168,603 "learning_rate": 4.962666666666667e-05,604 "loss": 1.6841,605 "step": 85606 },607 {608 "epoch": 0.007644444444444444,609 "grad_norm": 1.8857841491699219,610 "learning_rate": 4.9622222222222225e-05,611 "loss": 1.8312,612 "step": 86613 },614 {615 "epoch": 0.007733333333333333,616 "grad_norm": 2.4006853103637695,617 "learning_rate": 4.961777777777778e-05,618 "loss": 1.6762,619 "step": 87620 },621 {622 "epoch": 0.007822222222222222,623 "grad_norm": 2.2285337448120117,624 "learning_rate": 4.9613333333333335e-05,625 "loss": 1.8489,626 "step": 88627 },628 {629 "epoch": 0.007911111111111112,630 "grad_norm": 2.562854528427124,631 "learning_rate": 4.9608888888888897e-05,632 "loss": 1.968,633 "step": 89634 },635 {636 "epoch": 0.008,637 "grad_norm": 2.0626981258392334,638 "learning_rate": 4.9604444444444445e-05,639 "loss": 1.534,640 "step": 90641 },642 {643 "epoch": 0.008088888888888889,644 "grad_norm": 2.2159745693206787,645 "learning_rate": 4.96e-05,646 "loss": 1.8077,647 "step": 91648 },649 {650 "epoch": 0.008177777777777779,651 "grad_norm": 2.076690196990967,652 "learning_rate": 4.9595555555555554e-05,653 "loss": 1.8188,654 "step": 92655 },656 {657 "epoch": 0.008266666666666667,658 "grad_norm": 1.987101435661316,659 "learning_rate": 4.9591111111111116e-05,660 "loss": 1.5503,661 "step": 93662 },663 {664 "epoch": 0.008355555555555555,665 "grad_norm": 1.8614814281463623,666 "learning_rate": 4.958666666666667e-05,667 "loss": 1.6504,668 "step": 94669 },670 {671 "epoch": 0.008444444444444444,672 "grad_norm": 1.917844533920288,673 "learning_rate": 4.9582222222222226e-05,674 "loss": 1.7731,675 "step": 95676 },677 {678 "epoch": 0.008533333333333334,679 "grad_norm": 2.2739644050598145,680 "learning_rate": 4.957777777777778e-05,681 "loss": 1.746,682 "step": 96683 },684 {685 "epoch": 0.008622222222222222,686 "grad_norm": 2.8967883586883545,687 "learning_rate": 4.9573333333333335e-05,688 "loss": 1.8537,689 "step": 97690 },691 {692 "epoch": 0.00871111111111111,693 "grad_norm": 1.9918015003204346,694 "learning_rate": 4.956888888888889e-05,695 "loss": 1.4975,696 "step": 98697 },698 {699 "epoch": 0.0088,700 "grad_norm": 2.138683319091797,701 "learning_rate": 4.9564444444444445e-05,702 "loss": 1.884,703 "step": 99704 },705 {706 "epoch": 0.008888888888888889,707 "grad_norm": 1.9930676221847534,708 "learning_rate": 4.956e-05,709 "loss": 1.8581,710 "step": 100711 },712 {713 "epoch": 0.008977777777777777,714 "grad_norm": 2.752715826034546,715 "learning_rate": 4.955555555555556e-05,716 "loss": 1.9088,717 "step": 101718 },719 {720 "epoch": 0.009066666666666667,721 "grad_norm": 2.2268667221069336,722 "learning_rate": 4.9551111111111116e-05,723 "loss": 1.7784,724 "step": 102725 },726 {727 "epoch": 0.009155555555555556,728 "grad_norm": 2.253488063812256,729 "learning_rate": 4.954666666666667e-05,730 "loss": 1.7381,731 "step": 103732 },733 {734 "epoch": 0.009244444444444444,735 "grad_norm": 2.356184244155884,736 "learning_rate": 4.9542222222222226e-05,737 "loss": 1.9309,738 "step": 104739 },740 {741 "epoch": 0.009333333333333334,742 "grad_norm": 2.941348075866699,743 "learning_rate": 4.9537777777777774e-05,744 "loss": 1.9479,745 "step": 105746 },747 {748 "epoch": 0.009422222222222222,749 "grad_norm": 1.9343492984771729,750 "learning_rate": 4.9533333333333336e-05,751 "loss": 1.7302,752 "step": 106753 },754 {755 "epoch": 0.00951111111111111,756 "grad_norm": 2.6718897819519043,757 "learning_rate": 4.952888888888889e-05,758 "loss": 1.938,759 "step": 107760 },761 {762 "epoch": 0.0096,763 "grad_norm": 2.0558865070343018,764 "learning_rate": 4.9524444444444445e-05,765 "loss": 1.8748,766 "step": 108767 },768 {769 "epoch": 0.00968888888888889,770 "grad_norm": 2.030961275100708,771 "learning_rate": 4.952e-05,772 "loss": 1.7465,773 "step": 109774 },775 {776 "epoch": 0.009777777777777778,777 "grad_norm": 2.187312126159668,778 "learning_rate": 4.951555555555556e-05,779 "loss": 1.6758,780 "step": 110781 },782 {783 "epoch": 0.009866666666666666,784 "grad_norm": 2.035094976425171,785 "learning_rate": 4.951111111111112e-05,786 "loss": 1.7021,787 "step": 111788 },789 {790 "epoch": 0.009955555555555556,791 "grad_norm": 2.6232101917266846,792 "learning_rate": 4.9506666666666665e-05,793 "loss": 1.6382,794 "step": 112795 },796 {797 "epoch": 0.010044444444444444,798 "grad_norm": 2.269892692565918,799 "learning_rate": 4.950222222222222e-05,800 "loss": 1.7952,801 "step": 113802 },803 {804 "epoch": 0.010133333333333333,805 "grad_norm": 2.1355998516082764,806 "learning_rate": 4.949777777777778e-05,807 "loss": 1.7156,808 "step": 114809 },810 {811 "epoch": 0.010222222222222223,812 "grad_norm": 2.6449780464172363,813 "learning_rate": 4.9493333333333336e-05,814 "loss": 1.6516,815 "step": 115816 },817 {818 "epoch": 0.010311111111111111,819 "grad_norm": 3.118377923965454,820 "learning_rate": 4.948888888888889e-05,821 "loss": 1.6217,822 "step": 116823 },824 {825 "epoch": 0.0104,826 "grad_norm": 2.2711691856384277,827 "learning_rate": 4.9484444444444446e-05,828 "loss": 1.7195,829 "step": 117830 },831 {832 "epoch": 0.01048888888888889,833 "grad_norm": 2.1320762634277344,834 "learning_rate": 4.948000000000001e-05,835 "loss": 2.0947,836 "step": 118837 },838 {839 "epoch": 0.010577777777777778,840 "grad_norm": 2.022264242172241,841 "learning_rate": 4.9475555555555555e-05,842 "loss": 1.497,843 "step": 119844 },845 {846 "epoch": 0.010666666666666666,847 "grad_norm": 2.437167167663574,848 "learning_rate": 4.947111111111111e-05,849 "loss": 2.0855,850 "step": 120851 },852 {853 "epoch": 0.010755555555555556,854 "grad_norm": 1.9884405136108398,855 "learning_rate": 4.9466666666666665e-05,856 "loss": 1.9273,857 "step": 121858 },859 {860 "epoch": 0.010844444444444445,861 "grad_norm": 2.4299566745758057,862 "learning_rate": 4.946222222222223e-05,863 "loss": 1.8282,864 "step": 122865 },866 {867 "epoch": 0.010933333333333333,868 "grad_norm": 2.2237820625305176,869 "learning_rate": 4.945777777777778e-05,870 "loss": 1.783,871 "step": 123872 },873 {874 "epoch": 0.011022222222222221,875 "grad_norm": 2.0427815914154053,876 "learning_rate": 4.9453333333333336e-05,877 "loss": 1.8517,878 "step": 124879 },880 {881 "epoch": 0.011111111111111112,882 "grad_norm": 2.600771188735962,883 "learning_rate": 4.944888888888889e-05,884 "loss": 2.2527,885 "step": 125886 },887 {888 "epoch": 0.0112,889 "grad_norm": 2.1525046825408936,890 "learning_rate": 4.9444444444444446e-05,891 "loss": 1.7437,892 "step": 126893 },894 {895 "epoch": 0.011288888888888888,896 "grad_norm": 1.6396191120147705,897 "learning_rate": 4.944e-05,898 "loss": 1.7013,899 "step": 127900 },901 {902 "epoch": 0.011377777777777778,903 "grad_norm": 1.9528329372406006,904 "learning_rate": 4.9435555555555556e-05,905 "loss": 1.7621,906 "step": 128907 },908 {909 "epoch": 0.011466666666666667,910 "grad_norm": 2.3578553199768066,911 "learning_rate": 4.943111111111111e-05,912 "loss": 1.8193,913 "step": 129914 },915 {916 "epoch": 0.011555555555555555,917 "grad_norm": 2.2848048210144043,918 "learning_rate": 4.942666666666667e-05,919 "loss": 2.1061,920 "step": 130921 },922 {923 "epoch": 0.011644444444444445,924 "grad_norm": 2.1300034523010254,925 "learning_rate": 4.942222222222223e-05,926 "loss": 1.682,927 "step": 131928 },929 {930 "epoch": 0.011733333333333333,931 "grad_norm": 2.304337739944458,932 "learning_rate": 4.941777777777778e-05,933 "loss": 1.8597,934 "step": 132935 },936 {937 "epoch": 0.011822222222222222,938 "grad_norm": 2.2310221195220947,939 "learning_rate": 4.941333333333334e-05,940 "loss": 1.9288,941 "step": 133942 },943 {944 "epoch": 0.011911111111111112,945 "grad_norm": 1.9477310180664062,946 "learning_rate": 4.940888888888889e-05,947 "loss": 1.5615,948 "step": 134949 },950 {951 "epoch": 0.012,952 "grad_norm": 1.9804407358169556,953 "learning_rate": 4.9404444444444447e-05,954 "loss": 1.7701,955 "step": 135956 },957 {958 "epoch": 0.012088888888888889,959 "grad_norm": 1.9971662759780884,960 "learning_rate": 4.94e-05,961 "loss": 1.6494,962 "step": 136963 },964 {965 "epoch": 0.012177777777777777,966 "grad_norm": 2.1567671298980713,967 "learning_rate": 4.9395555555555556e-05,968 "loss": 1.8167,969 "step": 137970 },971 {972 "epoch": 0.012266666666666667,973 "grad_norm": 1.844053030014038,974 "learning_rate": 4.939111111111112e-05,975 "loss": 1.7869,976 "step": 138977 },978 {979 "epoch": 0.012355555555555555,980 "grad_norm": 2.9733023643493652,981 "learning_rate": 4.938666666666667e-05,982 "loss": 1.8699,983 "step": 139984 },985 {986 "epoch": 0.012444444444444444,987 "grad_norm": 2.2448322772979736,988 "learning_rate": 4.938222222222223e-05,989 "loss": 1.9895,990 "step": 140991 },992 {993 "epoch": 0.012533333333333334,994 "grad_norm": 2.020217180252075,995 "learning_rate": 4.9377777777777776e-05,996 "loss": 1.721,997 "step": 141998 },999 {1000 "epoch": 0.012622222222222222,1001 "grad_norm": 2.0747439861297607,1002 "learning_rate": 4.937333333333334e-05,1003 "loss": 1.6222,1004 "step": 1421005 },1006 {1007 "epoch": 0.01271111111111111,1008 "grad_norm": 2.166767120361328,1009 "learning_rate": 4.936888888888889e-05,1010 "loss": 1.6979,1011 "step": 1431012 },1013 {1014 "epoch": 0.0128,1015 "grad_norm": 2.1541872024536133,1016 "learning_rate": 4.936444444444445e-05,1017 "loss": 1.7211,1018 "step": 1441019 },1020 {1021 "epoch": 0.012888888888888889,1022 "grad_norm": 1.8028501272201538,1023 "learning_rate": 4.936e-05,1024 "loss": 1.6641,1025 "step": 1451026 },1027 {1028 "epoch": 0.012977777777777777,1029 "grad_norm": 1.9419965744018555,1030 "learning_rate": 4.935555555555556e-05,1031 "loss": 1.8437,1032 "step": 1461033 },1034 {1035 "epoch": 0.013066666666666667,1036 "grad_norm": 2.1536786556243896,1037 "learning_rate": 4.935111111111111e-05,1038 "loss": 1.6553,1039 "step": 1471040 },1041 {1042 "epoch": 0.013155555555555556,1043 "grad_norm": 2.270939350128174,1044 "learning_rate": 4.9346666666666666e-05,1045 "loss": 1.912,1046 "step": 1481047 },1048 {1049 "epoch": 0.013244444444444444,1050 "grad_norm": 4.608254909515381,1051 "learning_rate": 4.934222222222222e-05,1052 "loss": 2.3411,1053 "step": 1491054 },1055 {1056 "epoch": 0.013333333333333334,1057 "grad_norm": 2.3868587017059326,1058 "learning_rate": 4.933777777777778e-05,1059 "loss": 1.7206,1060 "step": 1501061 },1062 {1063 "epoch": 0.013422222222222223,1064 "grad_norm": 2.1859641075134277,1065 "learning_rate": 4.933333333333334e-05,1066 "loss": 1.7658,1067 "step": 1511068 },1069 {1070 "epoch": 0.013511111111111111,1071 "grad_norm": 2.1429860591888428,1072 "learning_rate": 4.932888888888889e-05,1073 "loss": 1.9059,1074 "step": 1521075 },1076 {1077 "epoch": 0.0136,1078 "grad_norm": 2.381554126739502,1079 "learning_rate": 4.932444444444445e-05,1080 "loss": 1.5929,1081 "step": 1531082 },1083 {1084 "epoch": 0.01368888888888889,1085 "grad_norm": 1.8531153202056885,1086 "learning_rate": 4.932e-05,1087 "loss": 1.8187,1088 "step": 1541089 },1090 {1091 "epoch": 0.013777777777777778,1092 "grad_norm": 2.0565645694732666,1093 "learning_rate": 4.931555555555556e-05,1094 "loss": 1.8435,1095 "step": 1551096 },1097 {1098 "epoch": 0.013866666666666666,1099 "grad_norm": 1.8367215394973755,1100 "learning_rate": 4.931111111111111e-05,1101 "loss": 1.6174,1102 "step": 1561103 },1104 {1105 "epoch": 0.013955555555555556,1106 "grad_norm": 2.1454102993011475,1107 "learning_rate": 4.930666666666667e-05,1108 "loss": 1.874,1109 "step": 1571110 },1111 {1112 "epoch": 0.014044444444444444,1113 "grad_norm": 2.2874836921691895,1114 "learning_rate": 4.930222222222222e-05,1115 "loss": 1.929,1116 "step": 1581117 },1118 {1119 "epoch": 0.014133333333333333,1120 "grad_norm": 1.7943480014801025,1121 "learning_rate": 4.929777777777778e-05,1122 "loss": 1.6639,1123 "step": 1591124 },1125 {1126 "epoch": 0.014222222222222223,1127 "grad_norm": 2.146174430847168,1128 "learning_rate": 4.929333333333334e-05,1129 "loss": 1.7037,1130 "step": 1601131 },1132 {1133 "epoch": 0.014311111111111111,1134 "grad_norm": 2.03281569480896,1135 "learning_rate": 4.928888888888889e-05,1136 "loss": 2.0383,1137 "step": 1611138 },1139 {1140 "epoch": 0.0144,1141 "grad_norm": 2.116950511932373,1142 "learning_rate": 4.928444444444444e-05,1143 "loss": 1.7421,1144 "step": 1621145 },1146 {1147 "epoch": 0.01448888888888889,1148 "grad_norm": 2.5180981159210205,1149 "learning_rate": 4.928e-05,1150 "loss": 2.052,1151 "step": 1631152 },1153 {1154 "epoch": 0.014577777777777778,1155 "grad_norm": 1.8855026960372925,1156 "learning_rate": 4.927555555555556e-05,1157 "loss": 1.6839,1158 "step": 1641159 },1160 {1161 "epoch": 0.014666666666666666,1162 "grad_norm": 3.8352456092834473,1163 "learning_rate": 4.927111111111111e-05,1164 "loss": 1.796,1165 "step": 1651166 },1167 {1168 "epoch": 0.014755555555555555,1169 "grad_norm": 2.5179617404937744,1170 "learning_rate": 4.926666666666667e-05,1171 "loss": 2.03,1172 "step": 1661173 },1174 {1175 "epoch": 0.014844444444444445,1176 "grad_norm": 1.9074064493179321,1177 "learning_rate": 4.926222222222223e-05,1178 "loss": 1.7468,1179 "step": 1671180 },1181 {1182 "epoch": 0.014933333333333333,1183 "grad_norm": 1.9903556108474731,1184 "learning_rate": 4.9257777777777784e-05,1185 "loss": 1.7349,1186 "step": 1681187 },1188 {1189 "epoch": 0.015022222222222222,1190 "grad_norm": 2.2186553478240967,1191 "learning_rate": 4.925333333333333e-05,1192 "loss": 1.8371,1193 "step": 1691194 },1195 {1196 "epoch": 0.015111111111111112,1197 "grad_norm": 2.206364154815674,1198 "learning_rate": 4.9248888888888886e-05,1199 "loss": 1.9397,1200 "step": 170