| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.007629016200215901, |
| "eval_steps": 200.0, |
| "global_step": 1000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 7.6290162002159015e-06, |
| "grad_norm": 19.858008765227513, |
| "learning_rate": 7.628928898382667e-10, |
| "loss": 0.6787815093994141, |
| "step": 1, |
| "token_acc": 0.8377192982456141 |
| }, |
| { |
| "epoch": 3.81450810010795e-05, |
| "grad_norm": 26.05024103541224, |
| "learning_rate": 3.814464449191334e-09, |
| "loss": 0.7937984466552734, |
| "step": 5, |
| "token_acc": 0.8203161105462612 |
| }, |
| { |
| "epoch": 7.6290162002159e-05, |
| "grad_norm": 30.045629754543953, |
| "learning_rate": 7.628928898382667e-09, |
| "loss": 0.722916555404663, |
| "step": 10, |
| "token_acc": 0.8339535988597672 |
| }, |
| { |
| "epoch": 0.00011443524300323851, |
| "grad_norm": 41.77187019490099, |
| "learning_rate": 1.1443393347574e-08, |
| "loss": 0.7019518852233887, |
| "step": 15, |
| "token_acc": 0.8500863778872608 |
| }, |
| { |
| "epoch": 0.000152580324004318, |
| "grad_norm": 13.031898089287347, |
| "learning_rate": 1.5257857796765334e-08, |
| "loss": 0.7568827629089355, |
| "step": 20, |
| "token_acc": 0.8252126836813611 |
| }, |
| { |
| "epoch": 0.00019072540500539753, |
| "grad_norm": 26.463184090371293, |
| "learning_rate": 1.907232224595667e-08, |
| "loss": 0.7129611968994141, |
| "step": 25, |
| "token_acc": 0.8335900354918636 |
| }, |
| { |
| "epoch": 0.00022887048600647703, |
| "grad_norm": 16.079871921811126, |
| "learning_rate": 2.2886786695148e-08, |
| "loss": 0.7088653564453125, |
| "step": 30, |
| "token_acc": 0.8355570839064649 |
| }, |
| { |
| "epoch": 0.0002670155670075565, |
| "grad_norm": 33.687562516956945, |
| "learning_rate": 2.6701251144339337e-08, |
| "loss": 0.7810484886169433, |
| "step": 35, |
| "token_acc": 0.8183418693982074 |
| }, |
| { |
| "epoch": 0.000305160648008636, |
| "grad_norm": 17.19646111197456, |
| "learning_rate": 3.051571559353067e-08, |
| "loss": 0.6474244117736816, |
| "step": 40, |
| "token_acc": 0.8510296480210311 |
| }, |
| { |
| "epoch": 0.00034330572900971557, |
| "grad_norm": 24.770380370056916, |
| "learning_rate": 3.433018004272201e-08, |
| "loss": 0.66593337059021, |
| "step": 45, |
| "token_acc": 0.8447153957925327 |
| }, |
| { |
| "epoch": 0.00038145081001079507, |
| "grad_norm": 14.223809744043987, |
| "learning_rate": 3.814464449191334e-08, |
| "loss": 0.6652703762054444, |
| "step": 50, |
| "token_acc": 0.8436879598271634 |
| }, |
| { |
| "epoch": 0.00041959589101187456, |
| "grad_norm": 19.829855717229282, |
| "learning_rate": 4.195910894110467e-08, |
| "loss": 0.6511832237243652, |
| "step": 55, |
| "token_acc": 0.8471829658941021 |
| }, |
| { |
| "epoch": 0.00045774097201295406, |
| "grad_norm": 15.85798445068367, |
| "learning_rate": 4.5773573390296e-08, |
| "loss": 0.6936227798461914, |
| "step": 60, |
| "token_acc": 0.8382894155748565 |
| }, |
| { |
| "epoch": 0.0004958860530140336, |
| "grad_norm": 19.984224693650123, |
| "learning_rate": 4.958803783948734e-08, |
| "loss": 0.6485838890075684, |
| "step": 65, |
| "token_acc": 0.8432868352223191 |
| }, |
| { |
| "epoch": 0.000534031134015113, |
| "grad_norm": 13.967273121899467, |
| "learning_rate": 5.3402502288678674e-08, |
| "loss": 0.655977439880371, |
| "step": 70, |
| "token_acc": 0.8395167678901505 |
| }, |
| { |
| "epoch": 0.0005721762150161925, |
| "grad_norm": 23.9354021403603, |
| "learning_rate": 5.7216966737870006e-08, |
| "loss": 0.7054988861083984, |
| "step": 75, |
| "token_acc": 0.8299424184261036 |
| }, |
| { |
| "epoch": 0.000610321296017272, |
| "grad_norm": 14.366371084466175, |
| "learning_rate": 6.103143118706134e-08, |
| "loss": 0.6200805187225342, |
| "step": 80, |
| "token_acc": 0.8412296564195298 |
| }, |
| { |
| "epoch": 0.0006484663770183516, |
| "grad_norm": 13.927963325668722, |
| "learning_rate": 6.484589563625268e-08, |
| "loss": 0.5447507858276367, |
| "step": 85, |
| "token_acc": 0.8580637086175605 |
| }, |
| { |
| "epoch": 0.0006866114580194311, |
| "grad_norm": 17.084378717846253, |
| "learning_rate": 6.866036008544401e-08, |
| "loss": 0.5300666809082031, |
| "step": 90, |
| "token_acc": 0.8586914189276905 |
| }, |
| { |
| "epoch": 0.0007247565390205106, |
| "grad_norm": 12.831399981259963, |
| "learning_rate": 7.247482453463534e-08, |
| "loss": 0.5949723720550537, |
| "step": 95, |
| "token_acc": 0.8456714481366684 |
| }, |
| { |
| "epoch": 0.0007629016200215901, |
| "grad_norm": 9.654604645266131, |
| "learning_rate": 7.628928898382668e-08, |
| "loss": 0.5178804397583008, |
| "step": 100, |
| "token_acc": 0.8577652050919378 |
| }, |
| { |
| "epoch": 0.0008010467010226696, |
| "grad_norm": 9.419771464141803, |
| "learning_rate": 8.0103753433018e-08, |
| "loss": 0.5627875328063965, |
| "step": 105, |
| "token_acc": 0.8502238502238503 |
| }, |
| { |
| "epoch": 0.0008391917820237491, |
| "grad_norm": 10.700536352923347, |
| "learning_rate": 8.391821788220934e-08, |
| "loss": 0.5639850616455078, |
| "step": 110, |
| "token_acc": 0.8486978643636607 |
| }, |
| { |
| "epoch": 0.0008773368630248286, |
| "grad_norm": 18.083522443391814, |
| "learning_rate": 8.773268233140068e-08, |
| "loss": 0.5201422691345214, |
| "step": 115, |
| "token_acc": 0.8569876409072389 |
| }, |
| { |
| "epoch": 0.0009154819440259081, |
| "grad_norm": 8.157978934525882, |
| "learning_rate": 9.1547146780592e-08, |
| "loss": 0.5080629348754883, |
| "step": 120, |
| "token_acc": 0.8670629304651383 |
| }, |
| { |
| "epoch": 0.0009536270250269876, |
| "grad_norm": 7.8633829016429715, |
| "learning_rate": 9.536161122978335e-08, |
| "loss": 0.47640438079833985, |
| "step": 125, |
| "token_acc": 0.8592945806842265 |
| }, |
| { |
| "epoch": 0.000991772106028067, |
| "grad_norm": 10.996375473342454, |
| "learning_rate": 9.917607567897468e-08, |
| "loss": 0.4695584297180176, |
| "step": 130, |
| "token_acc": 0.870125663946107 |
| }, |
| { |
| "epoch": 0.0010299171870291467, |
| "grad_norm": 9.48145780304072, |
| "learning_rate": 1.0299054012816601e-07, |
| "loss": 0.41779556274414065, |
| "step": 135, |
| "token_acc": 0.8781450100252247 |
| }, |
| { |
| "epoch": 0.001068062268030226, |
| "grad_norm": 7.285733898226943, |
| "learning_rate": 1.0680500457735735e-07, |
| "loss": 0.42159037590026854, |
| "step": 140, |
| "token_acc": 0.8773419773095624 |
| }, |
| { |
| "epoch": 0.0011062073490313057, |
| "grad_norm": 6.059780464247734, |
| "learning_rate": 1.1061946902654869e-07, |
| "loss": 0.40407743453979494, |
| "step": 145, |
| "token_acc": 0.8766230324871482 |
| }, |
| { |
| "epoch": 0.001144352430032385, |
| "grad_norm": 7.452093118966406, |
| "learning_rate": 1.1443393347574001e-07, |
| "loss": 0.4567539215087891, |
| "step": 150, |
| "token_acc": 0.866771030732471 |
| }, |
| { |
| "epoch": 0.0011824975110334647, |
| "grad_norm": 6.407966242309223, |
| "learning_rate": 1.1824839792493135e-07, |
| "loss": 0.4267581939697266, |
| "step": 155, |
| "token_acc": 0.8669217106162044 |
| }, |
| { |
| "epoch": 0.001220642592034544, |
| "grad_norm": 6.675792771068159, |
| "learning_rate": 1.2206286237412268e-07, |
| "loss": 0.4057455062866211, |
| "step": 160, |
| "token_acc": 0.8701632511288642 |
| }, |
| { |
| "epoch": 0.0012587876730356237, |
| "grad_norm": 5.152432896859598, |
| "learning_rate": 1.2587732682331401e-07, |
| "loss": 0.417818546295166, |
| "step": 165, |
| "token_acc": 0.8704265211239135 |
| }, |
| { |
| "epoch": 0.0012969327540367033, |
| "grad_norm": 7.574436681211629, |
| "learning_rate": 1.2969179127250535e-07, |
| "loss": 0.3847885847091675, |
| "step": 170, |
| "token_acc": 0.8806957211824615 |
| }, |
| { |
| "epoch": 0.0013350778350377827, |
| "grad_norm": 5.227824773349837, |
| "learning_rate": 1.335062557216967e-07, |
| "loss": 0.39106528759002684, |
| "step": 175, |
| "token_acc": 0.8794553166158197 |
| }, |
| { |
| "epoch": 0.0013732229160388623, |
| "grad_norm": 5.370348537465429, |
| "learning_rate": 1.3732072017088803e-07, |
| "loss": 0.40700087547302244, |
| "step": 180, |
| "token_acc": 0.8752121614640986 |
| }, |
| { |
| "epoch": 0.0014113679970399417, |
| "grad_norm": 3.899114838688643, |
| "learning_rate": 1.4113518462007934e-07, |
| "loss": 0.33117434978485105, |
| "step": 185, |
| "token_acc": 0.8873824197210509 |
| }, |
| { |
| "epoch": 0.0014495130780410213, |
| "grad_norm": 7.517334555705535, |
| "learning_rate": 1.4494964906927068e-07, |
| "loss": 0.3739504337310791, |
| "step": 190, |
| "token_acc": 0.8766823414950543 |
| }, |
| { |
| "epoch": 0.0014876581590421007, |
| "grad_norm": 3.715493293332821, |
| "learning_rate": 1.4876411351846202e-07, |
| "loss": 0.36161179542541505, |
| "step": 195, |
| "token_acc": 0.880270056740645 |
| }, |
| { |
| "epoch": 0.0015258032400431803, |
| "grad_norm": 3.2584704553113855, |
| "learning_rate": 1.5257857796765336e-07, |
| "loss": 0.3406350612640381, |
| "step": 200, |
| "token_acc": 0.8852274327559498 |
| }, |
| { |
| "epoch": 0.0015639483210442597, |
| "grad_norm": 3.0327317389406394, |
| "learning_rate": 1.563930424168447e-07, |
| "loss": 0.33145263195037844, |
| "step": 205, |
| "token_acc": 0.8919872447248796 |
| }, |
| { |
| "epoch": 0.0016020934020453393, |
| "grad_norm": 3.651520865012892, |
| "learning_rate": 1.60207506866036e-07, |
| "loss": 0.3220701217651367, |
| "step": 210, |
| "token_acc": 0.8937093275488069 |
| }, |
| { |
| "epoch": 0.0016402384830464186, |
| "grad_norm": 4.110264197154364, |
| "learning_rate": 1.6402197131522735e-07, |
| "loss": 0.34341437816619874, |
| "step": 215, |
| "token_acc": 0.8845404225461916 |
| }, |
| { |
| "epoch": 0.0016783835640474982, |
| "grad_norm": 3.673651186053855, |
| "learning_rate": 1.6783643576441869e-07, |
| "loss": 0.31849493980407717, |
| "step": 220, |
| "token_acc": 0.8914095343856587 |
| }, |
| { |
| "epoch": 0.0017165286450485779, |
| "grad_norm": 3.4701434837015306, |
| "learning_rate": 1.7165090021361002e-07, |
| "loss": 0.29149889945983887, |
| "step": 225, |
| "token_acc": 0.8938066697402797 |
| }, |
| { |
| "epoch": 0.0017546737260496572, |
| "grad_norm": 4.501326153361515, |
| "learning_rate": 1.7546536466280136e-07, |
| "loss": 0.2862607479095459, |
| "step": 230, |
| "token_acc": 0.9036284845740896 |
| }, |
| { |
| "epoch": 0.0017928188070507368, |
| "grad_norm": 3.8208490412219533, |
| "learning_rate": 1.792798291119927e-07, |
| "loss": 0.27682464122772216, |
| "step": 235, |
| "token_acc": 0.9005648948478867 |
| }, |
| { |
| "epoch": 0.0018309638880518162, |
| "grad_norm": 3.1051066643962617, |
| "learning_rate": 1.83094293561184e-07, |
| "loss": 0.30829055309295655, |
| "step": 240, |
| "token_acc": 0.8974551460100593 |
| }, |
| { |
| "epoch": 0.0018691089690528958, |
| "grad_norm": 2.6304367853356623, |
| "learning_rate": 1.8690875801037535e-07, |
| "loss": 0.2855536937713623, |
| "step": 245, |
| "token_acc": 0.8998719762819217 |
| }, |
| { |
| "epoch": 0.0019072540500539752, |
| "grad_norm": 3.0160339634063056, |
| "learning_rate": 1.907232224595667e-07, |
| "loss": 0.2821728467941284, |
| "step": 250, |
| "token_acc": 0.9010830871771567 |
| }, |
| { |
| "epoch": 0.0019453991310550548, |
| "grad_norm": 3.3623406713328077, |
| "learning_rate": 1.9453768690875803e-07, |
| "loss": 0.28752238750457765, |
| "step": 255, |
| "token_acc": 0.8981417756366139 |
| }, |
| { |
| "epoch": 0.001983544212056134, |
| "grad_norm": 2.858724195529313, |
| "learning_rate": 1.9835215135794937e-07, |
| "loss": 0.2832742691040039, |
| "step": 260, |
| "token_acc": 0.9023319106822539 |
| }, |
| { |
| "epoch": 0.0020216892930572136, |
| "grad_norm": 3.7239371602902143, |
| "learning_rate": 2.0216661580714068e-07, |
| "loss": 0.2549095869064331, |
| "step": 265, |
| "token_acc": 0.9071066362901448 |
| }, |
| { |
| "epoch": 0.0020598343740582934, |
| "grad_norm": 3.481525766253563, |
| "learning_rate": 2.0598108025633202e-07, |
| "loss": 0.30463471412658694, |
| "step": 270, |
| "token_acc": 0.8938873626373627 |
| }, |
| { |
| "epoch": 0.002097979455059373, |
| "grad_norm": 2.8701813267844356, |
| "learning_rate": 2.0979554470552336e-07, |
| "loss": 0.27567908763885496, |
| "step": 275, |
| "token_acc": 0.9044994143181975 |
| }, |
| { |
| "epoch": 0.002136124536060452, |
| "grad_norm": 2.1624912096505766, |
| "learning_rate": 2.136100091547147e-07, |
| "loss": 0.27143154144287107, |
| "step": 280, |
| "token_acc": 0.9036544850498339 |
| }, |
| { |
| "epoch": 0.002174269617061532, |
| "grad_norm": 4.728168119300998, |
| "learning_rate": 2.1742447360390603e-07, |
| "loss": 0.26180579662323, |
| "step": 285, |
| "token_acc": 0.9078116254897257 |
| }, |
| { |
| "epoch": 0.0022124146980626114, |
| "grad_norm": 3.0085172830041564, |
| "learning_rate": 2.2123893805309737e-07, |
| "loss": 0.27187042236328124, |
| "step": 290, |
| "token_acc": 0.9030493094014813 |
| }, |
| { |
| "epoch": 0.002250559779063691, |
| "grad_norm": 2.7990717081011027, |
| "learning_rate": 2.2505340250228868e-07, |
| "loss": 0.27347536087036134, |
| "step": 295, |
| "token_acc": 0.9003181985069147 |
| }, |
| { |
| "epoch": 0.00228870486006477, |
| "grad_norm": 3.2087776560137002, |
| "learning_rate": 2.2886786695148002e-07, |
| "loss": 0.2732613325119019, |
| "step": 300, |
| "token_acc": 0.9009136735979836 |
| }, |
| { |
| "epoch": 0.00232684994106585, |
| "grad_norm": 2.7855962147313456, |
| "learning_rate": 2.3268233140067136e-07, |
| "loss": 0.25121102333068845, |
| "step": 305, |
| "token_acc": 0.9103824400754107 |
| }, |
| { |
| "epoch": 0.0023649950220669294, |
| "grad_norm": 23.00066788095994, |
| "learning_rate": 2.364967958498627e-07, |
| "loss": 0.29934229850769045, |
| "step": 310, |
| "token_acc": 0.8942279519484324 |
| }, |
| { |
| "epoch": 0.0024031401030680088, |
| "grad_norm": 4.913378638209299, |
| "learning_rate": 2.40311260299054e-07, |
| "loss": 0.2760341644287109, |
| "step": 315, |
| "token_acc": 0.9069470572426767 |
| }, |
| { |
| "epoch": 0.002441285184069088, |
| "grad_norm": 3.7230064334029422, |
| "learning_rate": 2.4412572474824535e-07, |
| "loss": 0.316583514213562, |
| "step": 320, |
| "token_acc": 0.8880706921944035 |
| }, |
| { |
| "epoch": 0.002479430265070168, |
| "grad_norm": 3.0630673787948273, |
| "learning_rate": 2.479401891974367e-07, |
| "loss": 0.28699023723602296, |
| "step": 325, |
| "token_acc": 0.8988292744005509 |
| }, |
| { |
| "epoch": 0.0025175753460712474, |
| "grad_norm": 2.7791052705585235, |
| "learning_rate": 2.5175465364662803e-07, |
| "loss": 0.2764303207397461, |
| "step": 330, |
| "token_acc": 0.8995172244497177 |
| }, |
| { |
| "epoch": 0.0025557204270723268, |
| "grad_norm": 4.808529142134823, |
| "learning_rate": 2.5556911809581937e-07, |
| "loss": 0.27858314514160154, |
| "step": 335, |
| "token_acc": 0.9007542460252862 |
| }, |
| { |
| "epoch": 0.0025938655080734066, |
| "grad_norm": 2.646253308159165, |
| "learning_rate": 2.593835825450107e-07, |
| "loss": 0.25584604740142824, |
| "step": 340, |
| "token_acc": 0.9082719082719083 |
| }, |
| { |
| "epoch": 0.002632010589074486, |
| "grad_norm": 5.050994193531453, |
| "learning_rate": 2.6319804699420204e-07, |
| "loss": 0.3128068447113037, |
| "step": 345, |
| "token_acc": 0.8978692138133725 |
| }, |
| { |
| "epoch": 0.0026701556700755654, |
| "grad_norm": 2.922439667410607, |
| "learning_rate": 2.670125114433934e-07, |
| "loss": 0.2790635108947754, |
| "step": 350, |
| "token_acc": 0.9022727272727272 |
| }, |
| { |
| "epoch": 0.0027083007510766447, |
| "grad_norm": 2.606552490620242, |
| "learning_rate": 2.708269758925847e-07, |
| "loss": 0.22434093952178955, |
| "step": 355, |
| "token_acc": 0.9184911328057305 |
| }, |
| { |
| "epoch": 0.0027464458320777246, |
| "grad_norm": 2.2464414779398516, |
| "learning_rate": 2.7464144034177606e-07, |
| "loss": 0.2485731601715088, |
| "step": 360, |
| "token_acc": 0.913417007852876 |
| }, |
| { |
| "epoch": 0.002784590913078804, |
| "grad_norm": 2.232285266114994, |
| "learning_rate": 2.7845590479096734e-07, |
| "loss": 0.2663278579711914, |
| "step": 365, |
| "token_acc": 0.9048203868590727 |
| }, |
| { |
| "epoch": 0.0028227359940798833, |
| "grad_norm": 4.178002931261007, |
| "learning_rate": 2.822703692401587e-07, |
| "loss": 0.2515146255493164, |
| "step": 370, |
| "token_acc": 0.9104171998976197 |
| }, |
| { |
| "epoch": 0.0028608810750809627, |
| "grad_norm": 2.890332520635348, |
| "learning_rate": 2.8608483368935e-07, |
| "loss": 0.25286097526550294, |
| "step": 375, |
| "token_acc": 0.9091865357643759 |
| }, |
| { |
| "epoch": 0.0028990261560820425, |
| "grad_norm": 3.0083368175320047, |
| "learning_rate": 2.8989929813854136e-07, |
| "loss": 0.25886220932006837, |
| "step": 380, |
| "token_acc": 0.907826982492276 |
| }, |
| { |
| "epoch": 0.002937171237083122, |
| "grad_norm": 4.890289775308246, |
| "learning_rate": 2.937137625877327e-07, |
| "loss": 0.27796077728271484, |
| "step": 385, |
| "token_acc": 0.902413515687852 |
| }, |
| { |
| "epoch": 0.0029753163180842013, |
| "grad_norm": 2.4007652310291543, |
| "learning_rate": 2.9752822703692404e-07, |
| "loss": 0.22965149879455565, |
| "step": 390, |
| "token_acc": 0.9193692149468632 |
| }, |
| { |
| "epoch": 0.003013461399085281, |
| "grad_norm": 2.703034695380086, |
| "learning_rate": 3.013426914861154e-07, |
| "loss": 0.26880793571472167, |
| "step": 395, |
| "token_acc": 0.9011542497376706 |
| }, |
| { |
| "epoch": 0.0030516064800863605, |
| "grad_norm": 2.7471412083206705, |
| "learning_rate": 3.051571559353067e-07, |
| "loss": 0.22871689796447753, |
| "step": 400, |
| "token_acc": 0.9171862392723539 |
| }, |
| { |
| "epoch": 0.00308975156108744, |
| "grad_norm": 2.8979547332295907, |
| "learning_rate": 3.0897162038449805e-07, |
| "loss": 0.2612307071685791, |
| "step": 405, |
| "token_acc": 0.904684700059732 |
| }, |
| { |
| "epoch": 0.0031278966420885193, |
| "grad_norm": 2.656438254711223, |
| "learning_rate": 3.127860848336894e-07, |
| "loss": 0.25812110900878904, |
| "step": 410, |
| "token_acc": 0.9117283950617284 |
| }, |
| { |
| "epoch": 0.003166041723089599, |
| "grad_norm": 2.80054119033248, |
| "learning_rate": 3.1660054928288073e-07, |
| "loss": 0.25269036293029784, |
| "step": 415, |
| "token_acc": 0.907069209515466 |
| }, |
| { |
| "epoch": 0.0032041868040906785, |
| "grad_norm": 7.1585542884169495, |
| "learning_rate": 3.20415013732072e-07, |
| "loss": 0.25886504650115966, |
| "step": 420, |
| "token_acc": 0.9069200430261742 |
| }, |
| { |
| "epoch": 0.003242331885091758, |
| "grad_norm": 2.340069613916146, |
| "learning_rate": 3.242294781812634e-07, |
| "loss": 0.27744815349578855, |
| "step": 425, |
| "token_acc": 0.8992904554818036 |
| }, |
| { |
| "epoch": 0.0032804769660928373, |
| "grad_norm": 3.0643890118779264, |
| "learning_rate": 3.280439426304547e-07, |
| "loss": 0.2664801120758057, |
| "step": 430, |
| "token_acc": 0.9005569456776172 |
| }, |
| { |
| "epoch": 0.003318622047093917, |
| "grad_norm": 2.708739491897395, |
| "learning_rate": 3.318584070796461e-07, |
| "loss": 0.26904611587524413, |
| "step": 435, |
| "token_acc": 0.9000482082596818 |
| }, |
| { |
| "epoch": 0.0033567671280949965, |
| "grad_norm": 4.501313817843271, |
| "learning_rate": 3.3567287152883737e-07, |
| "loss": 0.25872418880462644, |
| "step": 440, |
| "token_acc": 0.9131074027787988 |
| }, |
| { |
| "epoch": 0.003394912209096076, |
| "grad_norm": 2.9891978108449226, |
| "learning_rate": 3.3948733597802876e-07, |
| "loss": 0.27869994640350343, |
| "step": 445, |
| "token_acc": 0.8993040035757615 |
| }, |
| { |
| "epoch": 0.0034330572900971557, |
| "grad_norm": 4.01992715262689, |
| "learning_rate": 3.4330180042722005e-07, |
| "loss": 0.24958600997924804, |
| "step": 450, |
| "token_acc": 0.9094727018198787 |
| }, |
| { |
| "epoch": 0.003471202371098235, |
| "grad_norm": 2.542148367897184, |
| "learning_rate": 3.471162648764114e-07, |
| "loss": 0.2729614734649658, |
| "step": 455, |
| "token_acc": 0.8986539984164688 |
| }, |
| { |
| "epoch": 0.0035093474520993145, |
| "grad_norm": 2.3102049515467575, |
| "learning_rate": 3.509307293256027e-07, |
| "loss": 0.25960521697998046, |
| "step": 460, |
| "token_acc": 0.9041880532619208 |
| }, |
| { |
| "epoch": 0.003547492533100394, |
| "grad_norm": 4.215860115991377, |
| "learning_rate": 3.5474519377479406e-07, |
| "loss": 0.2580498218536377, |
| "step": 465, |
| "token_acc": 0.908362295198241 |
| }, |
| { |
| "epoch": 0.0035856376141014737, |
| "grad_norm": 2.5873046950881164, |
| "learning_rate": 3.585596582239854e-07, |
| "loss": 0.25073535442352296, |
| "step": 470, |
| "token_acc": 0.9081341984491217 |
| }, |
| { |
| "epoch": 0.003623782695102553, |
| "grad_norm": 9.841945468118112, |
| "learning_rate": 3.6237412267317674e-07, |
| "loss": 0.2642046928405762, |
| "step": 475, |
| "token_acc": 0.9030493273542601 |
| }, |
| { |
| "epoch": 0.0036619277761036325, |
| "grad_norm": 6.882617587596199, |
| "learning_rate": 3.66188587122368e-07, |
| "loss": 0.23246357440948487, |
| "step": 480, |
| "token_acc": 0.9140461215932913 |
| }, |
| { |
| "epoch": 0.003700072857104712, |
| "grad_norm": 2.0635484069380623, |
| "learning_rate": 3.700030515715594e-07, |
| "loss": 0.23335707187652588, |
| "step": 485, |
| "token_acc": 0.9135682885030921 |
| }, |
| { |
| "epoch": 0.0037382179381057917, |
| "grad_norm": 3.4740093175635174, |
| "learning_rate": 3.738175160207507e-07, |
| "loss": 0.2697558879852295, |
| "step": 490, |
| "token_acc": 0.9013763271726307 |
| }, |
| { |
| "epoch": 0.003776363019106871, |
| "grad_norm": 3.431739461613957, |
| "learning_rate": 3.776319804699421e-07, |
| "loss": 0.2169642448425293, |
| "step": 495, |
| "token_acc": 0.919410362785771 |
| }, |
| { |
| "epoch": 0.0038145081001079504, |
| "grad_norm": 1.5264757240403999, |
| "learning_rate": 3.814464449191334e-07, |
| "loss": 0.25090970993041994, |
| "step": 500, |
| "token_acc": 0.9121357432681667 |
| }, |
| { |
| "epoch": 0.0038526531811090303, |
| "grad_norm": 2.1883666824963797, |
| "learning_rate": 3.8526090936832477e-07, |
| "loss": 0.2483682632446289, |
| "step": 505, |
| "token_acc": 0.9091301932218038 |
| }, |
| { |
| "epoch": 0.0038907982621101097, |
| "grad_norm": 2.083496301448179, |
| "learning_rate": 3.8907537381751606e-07, |
| "loss": 0.2711325645446777, |
| "step": 510, |
| "token_acc": 0.9028123626776037 |
| }, |
| { |
| "epoch": 0.0039289433431111895, |
| "grad_norm": 2.641802435876891, |
| "learning_rate": 3.928898382667074e-07, |
| "loss": 0.27882430553436277, |
| "step": 515, |
| "token_acc": 0.9007821552723059 |
| }, |
| { |
| "epoch": 0.003967088424112268, |
| "grad_norm": 2.394684311531086, |
| "learning_rate": 3.9670430271589873e-07, |
| "loss": 0.2641146183013916, |
| "step": 520, |
| "token_acc": 0.9018478727975935 |
| }, |
| { |
| "epoch": 0.004005233505113348, |
| "grad_norm": 2.481739144309257, |
| "learning_rate": 4.0051876716509007e-07, |
| "loss": 0.26821279525756836, |
| "step": 525, |
| "token_acc": 0.9070000771783592 |
| }, |
| { |
| "epoch": 0.004043378586114427, |
| "grad_norm": 3.889296103853822, |
| "learning_rate": 4.0433323161428136e-07, |
| "loss": 0.22691609859466552, |
| "step": 530, |
| "token_acc": 0.9154143728611813 |
| }, |
| { |
| "epoch": 0.004081523667115507, |
| "grad_norm": 2.6368865169230022, |
| "learning_rate": 4.0814769606347275e-07, |
| "loss": 0.2556874752044678, |
| "step": 535, |
| "token_acc": 0.9065021725636251 |
| }, |
| { |
| "epoch": 0.004119668748116587, |
| "grad_norm": 2.6155085833020206, |
| "learning_rate": 4.1196216051266404e-07, |
| "loss": 0.21920876502990722, |
| "step": 540, |
| "token_acc": 0.92066594281578 |
| }, |
| { |
| "epoch": 0.004157813829117666, |
| "grad_norm": 1.9328229536953851, |
| "learning_rate": 4.1577662496185543e-07, |
| "loss": 0.23413827419281005, |
| "step": 545, |
| "token_acc": 0.909433962264151 |
| }, |
| { |
| "epoch": 0.004195958910118746, |
| "grad_norm": 2.9117172838218615, |
| "learning_rate": 4.195910894110467e-07, |
| "loss": 0.24360220432281493, |
| "step": 550, |
| "token_acc": 0.9103947742118716 |
| }, |
| { |
| "epoch": 0.0042341039911198254, |
| "grad_norm": 2.1250080249331256, |
| "learning_rate": 4.234055538602381e-07, |
| "loss": 0.24965639114379884, |
| "step": 555, |
| "token_acc": 0.9056687992858206 |
| }, |
| { |
| "epoch": 0.004272249072120904, |
| "grad_norm": 2.546842077158916, |
| "learning_rate": 4.272200183094294e-07, |
| "loss": 0.25338068008422854, |
| "step": 560, |
| "token_acc": 0.9083823425210799 |
| }, |
| { |
| "epoch": 0.004310394153121984, |
| "grad_norm": 2.055493648942245, |
| "learning_rate": 4.3103448275862073e-07, |
| "loss": 0.2616416931152344, |
| "step": 565, |
| "token_acc": 0.9030194104960461 |
| }, |
| { |
| "epoch": 0.004348539234123064, |
| "grad_norm": 3.081547453069715, |
| "learning_rate": 4.3484894720781207e-07, |
| "loss": 0.3007126569747925, |
| "step": 570, |
| "token_acc": 0.8967875647668394 |
| }, |
| { |
| "epoch": 0.004386684315124143, |
| "grad_norm": 2.6310833763661448, |
| "learning_rate": 4.386634116570034e-07, |
| "loss": 0.26825015544891356, |
| "step": 575, |
| "token_acc": 0.8971716563863397 |
| }, |
| { |
| "epoch": 0.004424829396125223, |
| "grad_norm": 1.9747613572324976, |
| "learning_rate": 4.4247787610619474e-07, |
| "loss": 0.24232356548309325, |
| "step": 580, |
| "token_acc": 0.9104139104139104 |
| }, |
| { |
| "epoch": 0.004462974477126302, |
| "grad_norm": 2.9507299451994164, |
| "learning_rate": 4.462923405553861e-07, |
| "loss": 0.2625190019607544, |
| "step": 585, |
| "token_acc": 0.9043351131451818 |
| }, |
| { |
| "epoch": 0.004501119558127382, |
| "grad_norm": 2.041702229587946, |
| "learning_rate": 4.5010680500457737e-07, |
| "loss": 0.27087769508361814, |
| "step": 590, |
| "token_acc": 0.8977926041592732 |
| }, |
| { |
| "epoch": 0.004539264639128461, |
| "grad_norm": 2.0495563743442253, |
| "learning_rate": 4.5392126945376876e-07, |
| "loss": 0.23960542678833008, |
| "step": 595, |
| "token_acc": 0.9126592532587083 |
| }, |
| { |
| "epoch": 0.00457740972012954, |
| "grad_norm": 2.436198073944371, |
| "learning_rate": 4.5773573390296005e-07, |
| "loss": 0.24157366752624512, |
| "step": 600, |
| "token_acc": 0.9103791174642635 |
| }, |
| { |
| "epoch": 0.00461555480113062, |
| "grad_norm": 2.5317430593867423, |
| "learning_rate": 4.6155019835215144e-07, |
| "loss": 0.271079421043396, |
| "step": 605, |
| "token_acc": 0.899757139268068 |
| }, |
| { |
| "epoch": 0.0046536998821317, |
| "grad_norm": 2.5553307431531627, |
| "learning_rate": 4.653646628013427e-07, |
| "loss": 0.24459531307220458, |
| "step": 610, |
| "token_acc": 0.9086593883501729 |
| }, |
| { |
| "epoch": 0.004691844963132779, |
| "grad_norm": 2.8304214669453165, |
| "learning_rate": 4.691791272505341e-07, |
| "loss": 0.2665693283081055, |
| "step": 615, |
| "token_acc": 0.9054301403294692 |
| }, |
| { |
| "epoch": 0.004729990044133859, |
| "grad_norm": 2.0461521335651165, |
| "learning_rate": 4.729935916997254e-07, |
| "loss": 0.2225412130355835, |
| "step": 620, |
| "token_acc": 0.9145840584754612 |
| }, |
| { |
| "epoch": 0.004768135125134939, |
| "grad_norm": 2.2406641298382355, |
| "learning_rate": 4.7680805614891674e-07, |
| "loss": 0.22277250289916992, |
| "step": 625, |
| "token_acc": 0.9159432870370371 |
| }, |
| { |
| "epoch": 0.0048062802061360176, |
| "grad_norm": 2.546827343906107, |
| "learning_rate": 4.80622520598108e-07, |
| "loss": 0.1908212661743164, |
| "step": 630, |
| "token_acc": 0.9257950530035336 |
| }, |
| { |
| "epoch": 0.004844425287137097, |
| "grad_norm": 2.1841664235416802, |
| "learning_rate": 4.844369850472995e-07, |
| "loss": 0.25454857349395754, |
| "step": 635, |
| "token_acc": 0.9049563764618526 |
| }, |
| { |
| "epoch": 0.004882570368138176, |
| "grad_norm": 2.572298033816989, |
| "learning_rate": 4.882514494964907e-07, |
| "loss": 0.23939483165740966, |
| "step": 640, |
| "token_acc": 0.9100228579344739 |
| }, |
| { |
| "epoch": 0.004920715449139256, |
| "grad_norm": 3.4160927766057685, |
| "learning_rate": 4.920659139456821e-07, |
| "loss": 0.25036001205444336, |
| "step": 645, |
| "token_acc": 0.9049896907216495 |
| }, |
| { |
| "epoch": 0.004958860530140336, |
| "grad_norm": 3.1113323822613976, |
| "learning_rate": 4.958803783948734e-07, |
| "loss": 0.22428619861602783, |
| "step": 650, |
| "token_acc": 0.9159893601300428 |
| }, |
| { |
| "epoch": 0.004997005611141415, |
| "grad_norm": 1.8929422066777826, |
| "learning_rate": 4.996948428440647e-07, |
| "loss": 0.2436244249343872, |
| "step": 655, |
| "token_acc": 0.9110026722435227 |
| }, |
| { |
| "epoch": 0.005035150692142495, |
| "grad_norm": 2.311586558615121, |
| "learning_rate": 5.035093072932561e-07, |
| "loss": 0.2434394359588623, |
| "step": 660, |
| "token_acc": 0.9068422934795566 |
| }, |
| { |
| "epoch": 0.005073295773143575, |
| "grad_norm": 1.9488749634400302, |
| "learning_rate": 5.073237717424474e-07, |
| "loss": 0.22594902515411378, |
| "step": 665, |
| "token_acc": 0.9147198366464476 |
| }, |
| { |
| "epoch": 0.0051114408541446535, |
| "grad_norm": 3.571265015516316, |
| "learning_rate": 5.111382361916387e-07, |
| "loss": 0.22313566207885743, |
| "step": 670, |
| "token_acc": 0.9155100735529261 |
| }, |
| { |
| "epoch": 0.005149585935145733, |
| "grad_norm": 8.81853964601659, |
| "learning_rate": 5.149527006408301e-07, |
| "loss": 0.23009917736053467, |
| "step": 675, |
| "token_acc": 0.91564106558654 |
| }, |
| { |
| "epoch": 0.005187731016146813, |
| "grad_norm": 2.3832019428324736, |
| "learning_rate": 5.187671650900214e-07, |
| "loss": 0.2520387411117554, |
| "step": 680, |
| "token_acc": 0.9045643153526971 |
| }, |
| { |
| "epoch": 0.005225876097147892, |
| "grad_norm": 2.773453447881878, |
| "learning_rate": 5.225816295392127e-07, |
| "loss": 0.22851839065551757, |
| "step": 685, |
| "token_acc": 0.9147792706333974 |
| }, |
| { |
| "epoch": 0.005264021178148972, |
| "grad_norm": 2.188161059769218, |
| "learning_rate": 5.263960939884041e-07, |
| "loss": 0.21708831787109376, |
| "step": 690, |
| "token_acc": 0.9175445873081709 |
| }, |
| { |
| "epoch": 0.005302166259150051, |
| "grad_norm": 2.120373057447545, |
| "learning_rate": 5.302105584375954e-07, |
| "loss": 0.240578031539917, |
| "step": 695, |
| "token_acc": 0.9101432498914773 |
| }, |
| { |
| "epoch": 0.005340311340151131, |
| "grad_norm": 7.143314253272402, |
| "learning_rate": 5.340250228867868e-07, |
| "loss": 0.24711151123046876, |
| "step": 700, |
| "token_acc": 0.905367231638418 |
| }, |
| { |
| "epoch": 0.0053784564211522105, |
| "grad_norm": 2.509437280811482, |
| "learning_rate": 5.378394873359781e-07, |
| "loss": 0.2288766860961914, |
| "step": 705, |
| "token_acc": 0.9190667983889353 |
| }, |
| { |
| "epoch": 0.0054166015021532895, |
| "grad_norm": 4.435980751184006, |
| "learning_rate": 5.416539517851694e-07, |
| "loss": 0.24404430389404297, |
| "step": 710, |
| "token_acc": 0.91390625 |
| }, |
| { |
| "epoch": 0.005454746583154369, |
| "grad_norm": 2.1864120237187454, |
| "learning_rate": 5.454684162343608e-07, |
| "loss": 0.22964107990264893, |
| "step": 715, |
| "token_acc": 0.9149337004564887 |
| }, |
| { |
| "epoch": 0.005492891664155449, |
| "grad_norm": 2.168263736034898, |
| "learning_rate": 5.492828806835521e-07, |
| "loss": 0.24856891632080078, |
| "step": 720, |
| "token_acc": 0.9097530347425701 |
| }, |
| { |
| "epoch": 0.005531036745156528, |
| "grad_norm": 2.9088664787654213, |
| "learning_rate": 5.530973451327435e-07, |
| "loss": 0.2512509346008301, |
| "step": 725, |
| "token_acc": 0.9060345353590823 |
| }, |
| { |
| "epoch": 0.005569181826157608, |
| "grad_norm": 3.3320808612830874, |
| "learning_rate": 5.569118095819347e-07, |
| "loss": 0.2525000810623169, |
| "step": 730, |
| "token_acc": 0.9031252030407381 |
| }, |
| { |
| "epoch": 0.005607326907158688, |
| "grad_norm": 1.6179035871243548, |
| "learning_rate": 5.607262740311261e-07, |
| "loss": 0.2348179340362549, |
| "step": 735, |
| "token_acc": 0.9133819951338199 |
| }, |
| { |
| "epoch": 0.005645471988159767, |
| "grad_norm": 2.85667421103228, |
| "learning_rate": 5.645407384803174e-07, |
| "loss": 0.23236339092254638, |
| "step": 740, |
| "token_acc": 0.914976065648507 |
| }, |
| { |
| "epoch": 0.0056836170691608465, |
| "grad_norm": 2.6909733482462914, |
| "learning_rate": 5.683552029295088e-07, |
| "loss": 0.23209877014160157, |
| "step": 745, |
| "token_acc": 0.9132922854038392 |
| }, |
| { |
| "epoch": 0.0057217621501619255, |
| "grad_norm": 1.5951096324195422, |
| "learning_rate": 5.721696673787e-07, |
| "loss": 0.22777156829833983, |
| "step": 750, |
| "token_acc": 0.9133550488599349 |
| }, |
| { |
| "epoch": 0.005759907231163005, |
| "grad_norm": 2.3881038308988383, |
| "learning_rate": 5.759841318278915e-07, |
| "loss": 0.2042149543762207, |
| "step": 755, |
| "token_acc": 0.9194363390506014 |
| }, |
| { |
| "epoch": 0.005798052312164085, |
| "grad_norm": 3.325840053407865, |
| "learning_rate": 5.797985962770827e-07, |
| "loss": 0.24493722915649413, |
| "step": 760, |
| "token_acc": 0.9089112637621154 |
| }, |
| { |
| "epoch": 0.005836197393165164, |
| "grad_norm": 1.8938911522743977, |
| "learning_rate": 5.836130607262741e-07, |
| "loss": 0.21641106605529786, |
| "step": 765, |
| "token_acc": 0.9197261978842564 |
| }, |
| { |
| "epoch": 0.005874342474166244, |
| "grad_norm": 1.988679833448486, |
| "learning_rate": 5.874275251754654e-07, |
| "loss": 0.21464207172393798, |
| "step": 770, |
| "token_acc": 0.9175137584978957 |
| }, |
| { |
| "epoch": 0.005912487555167324, |
| "grad_norm": 1.8627516018080663, |
| "learning_rate": 5.912419896246567e-07, |
| "loss": 0.2498110294342041, |
| "step": 775, |
| "token_acc": 0.9056705670567057 |
| }, |
| { |
| "epoch": 0.005950632636168403, |
| "grad_norm": 1.2851350756074684, |
| "learning_rate": 5.950564540738481e-07, |
| "loss": 0.242911434173584, |
| "step": 780, |
| "token_acc": 0.9100091183278389 |
| }, |
| { |
| "epoch": 0.0059887777171694825, |
| "grad_norm": 1.7215295791488319, |
| "learning_rate": 5.988709185230394e-07, |
| "loss": 0.19092849493026734, |
| "step": 785, |
| "token_acc": 0.9254417816509319 |
| }, |
| { |
| "epoch": 0.006026922798170562, |
| "grad_norm": 2.0995252488631713, |
| "learning_rate": 6.026853829722308e-07, |
| "loss": 0.21847922801971437, |
| "step": 790, |
| "token_acc": 0.9216517010244087 |
| }, |
| { |
| "epoch": 0.006065067879171641, |
| "grad_norm": 2.814177734711112, |
| "learning_rate": 6.064998474214221e-07, |
| "loss": 0.24257607460021974, |
| "step": 795, |
| "token_acc": 0.9114211550940947 |
| }, |
| { |
| "epoch": 0.006103212960172721, |
| "grad_norm": 2.2026833559693153, |
| "learning_rate": 6.103143118706134e-07, |
| "loss": 0.23102183341979982, |
| "step": 800, |
| "token_acc": 0.9107298271789134 |
| }, |
| { |
| "epoch": 0.0061413580411738, |
| "grad_norm": 2.0108750944280174, |
| "learning_rate": 6.141287763198048e-07, |
| "loss": 0.24538402557373046, |
| "step": 805, |
| "token_acc": 0.9061550420464984 |
| }, |
| { |
| "epoch": 0.00617950312217488, |
| "grad_norm": 2.304744993055787, |
| "learning_rate": 6.179432407689961e-07, |
| "loss": 0.2557771921157837, |
| "step": 810, |
| "token_acc": 0.9081625797783003 |
| }, |
| { |
| "epoch": 0.00621764820317596, |
| "grad_norm": 8.924798548774792, |
| "learning_rate": 6.217577052181874e-07, |
| "loss": 0.24581964015960694, |
| "step": 815, |
| "token_acc": 0.90890815429074 |
| }, |
| { |
| "epoch": 0.006255793284177039, |
| "grad_norm": 1.4409669493856745, |
| "learning_rate": 6.255721696673788e-07, |
| "loss": 0.20607192516326905, |
| "step": 820, |
| "token_acc": 0.9191860465116279 |
| }, |
| { |
| "epoch": 0.0062939383651781184, |
| "grad_norm": 2.925558681675864, |
| "learning_rate": 6.293866341165701e-07, |
| "loss": 0.2149512767791748, |
| "step": 825, |
| "token_acc": 0.9190668029583309 |
| }, |
| { |
| "epoch": 0.006332083446179198, |
| "grad_norm": 1.7110063722551234, |
| "learning_rate": 6.332010985657615e-07, |
| "loss": 0.21946215629577637, |
| "step": 830, |
| "token_acc": 0.9161647628937348 |
| }, |
| { |
| "epoch": 0.006370228527180277, |
| "grad_norm": 2.444989400700814, |
| "learning_rate": 6.370155630149527e-07, |
| "loss": 0.21233253479003905, |
| "step": 835, |
| "token_acc": 0.9200024520321216 |
| }, |
| { |
| "epoch": 0.006408373608181357, |
| "grad_norm": 3.2838014012433985, |
| "learning_rate": 6.40830027464144e-07, |
| "loss": 0.22280681133270264, |
| "step": 840, |
| "token_acc": 0.9153393695063116 |
| }, |
| { |
| "epoch": 0.006446518689182437, |
| "grad_norm": 1.9215417073538057, |
| "learning_rate": 6.446444919133355e-07, |
| "loss": 0.2385124683380127, |
| "step": 845, |
| "token_acc": 0.9088968579234973 |
| }, |
| { |
| "epoch": 0.006484663770183516, |
| "grad_norm": 2.319220162643885, |
| "learning_rate": 6.484589563625268e-07, |
| "loss": 0.2364635944366455, |
| "step": 850, |
| "token_acc": 0.909977141320425 |
| }, |
| { |
| "epoch": 0.006522808851184596, |
| "grad_norm": 1.9237714003957875, |
| "learning_rate": 6.52273420811718e-07, |
| "loss": 0.23012151718139648, |
| "step": 855, |
| "token_acc": 0.9148593082318793 |
| }, |
| { |
| "epoch": 0.006560953932185675, |
| "grad_norm": 2.046769328041715, |
| "learning_rate": 6.560878852609094e-07, |
| "loss": 0.2327641487121582, |
| "step": 860, |
| "token_acc": 0.9113459879206213 |
| }, |
| { |
| "epoch": 0.006599099013186754, |
| "grad_norm": 1.6115036705671675, |
| "learning_rate": 6.599023497101008e-07, |
| "loss": 0.2237847089767456, |
| "step": 865, |
| "token_acc": 0.9167464114832536 |
| }, |
| { |
| "epoch": 0.006637244094187834, |
| "grad_norm": 3.3051233128989015, |
| "learning_rate": 6.637168141592922e-07, |
| "loss": 0.19322073459625244, |
| "step": 870, |
| "token_acc": 0.9275590551181102 |
| }, |
| { |
| "epoch": 0.006675389175188913, |
| "grad_norm": 2.4262761074290506, |
| "learning_rate": 6.675312786084834e-07, |
| "loss": 0.22160861492156983, |
| "step": 875, |
| "token_acc": 0.9151369669584863 |
| }, |
| { |
| "epoch": 0.006713534256189993, |
| "grad_norm": 1.9677832165578242, |
| "learning_rate": 6.713457430576747e-07, |
| "loss": 0.225478196144104, |
| "step": 880, |
| "token_acc": 0.9115168539325843 |
| }, |
| { |
| "epoch": 0.006751679337191073, |
| "grad_norm": 5.096878328776845, |
| "learning_rate": 6.751602075068661e-07, |
| "loss": 0.24097964763641358, |
| "step": 885, |
| "token_acc": 0.9146459747817652 |
| }, |
| { |
| "epoch": 0.006789824418192152, |
| "grad_norm": 1.688691754760285, |
| "learning_rate": 6.789746719560575e-07, |
| "loss": 0.2104334592819214, |
| "step": 890, |
| "token_acc": 0.9222782379066139 |
| }, |
| { |
| "epoch": 0.006827969499193232, |
| "grad_norm": 3.120436618558742, |
| "learning_rate": 6.827891364052488e-07, |
| "loss": 0.23382256031036378, |
| "step": 895, |
| "token_acc": 0.9112284069097889 |
| }, |
| { |
| "epoch": 0.006866114580194311, |
| "grad_norm": 2.2142725887946937, |
| "learning_rate": 6.866036008544401e-07, |
| "loss": 0.25021958351135254, |
| "step": 900, |
| "token_acc": 0.9044054747647562 |
| }, |
| { |
| "epoch": 0.00690425966119539, |
| "grad_norm": 4.399873851436453, |
| "learning_rate": 6.904180653036314e-07, |
| "loss": 0.2350818157196045, |
| "step": 905, |
| "token_acc": 0.9120438237790294 |
| }, |
| { |
| "epoch": 0.00694240474219647, |
| "grad_norm": 2.5238462069718905, |
| "learning_rate": 6.942325297528228e-07, |
| "loss": 0.21383306980133057, |
| "step": 910, |
| "token_acc": 0.9169596314704813 |
| }, |
| { |
| "epoch": 0.006980549823197549, |
| "grad_norm": 1.8196001493704197, |
| "learning_rate": 6.98046994202014e-07, |
| "loss": 0.20101499557495117, |
| "step": 915, |
| "token_acc": 0.9233510235026535 |
| }, |
| { |
| "epoch": 0.007018694904198629, |
| "grad_norm": 1.8331955196154692, |
| "learning_rate": 7.018614586512054e-07, |
| "loss": 0.2478102207183838, |
| "step": 920, |
| "token_acc": 0.9081678497618834 |
| }, |
| { |
| "epoch": 0.007056839985199709, |
| "grad_norm": 1.6893601989964881, |
| "learning_rate": 7.056759231003968e-07, |
| "loss": 0.2171755313873291, |
| "step": 925, |
| "token_acc": 0.9169847990560145 |
| }, |
| { |
| "epoch": 0.007094985066200788, |
| "grad_norm": 2.7750882619687682, |
| "learning_rate": 7.094903875495881e-07, |
| "loss": 0.2517444610595703, |
| "step": 930, |
| "token_acc": 0.905949540952794 |
| }, |
| { |
| "epoch": 0.0071331301472018676, |
| "grad_norm": 2.295636343184957, |
| "learning_rate": 7.133048519987794e-07, |
| "loss": 0.21493184566497803, |
| "step": 935, |
| "token_acc": 0.919338959212377 |
| }, |
| { |
| "epoch": 0.007171275228202947, |
| "grad_norm": 5.274781334339023, |
| "learning_rate": 7.171193164479708e-07, |
| "loss": 0.2255638360977173, |
| "step": 940, |
| "token_acc": 0.9191055958791852 |
| }, |
| { |
| "epoch": 0.007209420309204026, |
| "grad_norm": 4.249507608934204, |
| "learning_rate": 7.209337808971621e-07, |
| "loss": 0.23890252113342286, |
| "step": 945, |
| "token_acc": 0.9082201572551822 |
| }, |
| { |
| "epoch": 0.007247565390205106, |
| "grad_norm": 1.8318728175930554, |
| "learning_rate": 7.247482453463535e-07, |
| "loss": 0.23565545082092285, |
| "step": 950, |
| "token_acc": 0.9120185682164358 |
| }, |
| { |
| "epoch": 0.007285710471206186, |
| "grad_norm": 2.3153771765190125, |
| "learning_rate": 7.285627097955447e-07, |
| "loss": 0.22924485206604003, |
| "step": 955, |
| "token_acc": 0.9118163094057518 |
| }, |
| { |
| "epoch": 0.007323855552207265, |
| "grad_norm": 2.1444734209902667, |
| "learning_rate": 7.32377174244736e-07, |
| "loss": 0.2102113723754883, |
| "step": 960, |
| "token_acc": 0.9237033861980283 |
| }, |
| { |
| "epoch": 0.007362000633208345, |
| "grad_norm": 2.313230627888173, |
| "learning_rate": 7.361916386939275e-07, |
| "loss": 0.22157104015350343, |
| "step": 965, |
| "token_acc": 0.9158624102944816 |
| }, |
| { |
| "epoch": 0.007400145714209424, |
| "grad_norm": 1.8991530777901926, |
| "learning_rate": 7.400061031431188e-07, |
| "loss": 0.21107559204101561, |
| "step": 970, |
| "token_acc": 0.9221950473861205 |
| }, |
| { |
| "epoch": 0.0074382907952105035, |
| "grad_norm": 1.9998413654401805, |
| "learning_rate": 7.438205675923101e-07, |
| "loss": 0.21765711307525634, |
| "step": 975, |
| "token_acc": 0.9134157944814463 |
| }, |
| { |
| "epoch": 0.007476435876211583, |
| "grad_norm": 2.3376032357783734, |
| "learning_rate": 7.476350320415014e-07, |
| "loss": 0.22147510051727295, |
| "step": 980, |
| "token_acc": 0.9163215117784106 |
| }, |
| { |
| "epoch": 0.007514580957212662, |
| "grad_norm": 1.5533529464193574, |
| "learning_rate": 7.514494964906927e-07, |
| "loss": 0.1793646216392517, |
| "step": 985, |
| "token_acc": 0.9294284173631706 |
| }, |
| { |
| "epoch": 0.007552726038213742, |
| "grad_norm": 2.325698269527801, |
| "learning_rate": 7.552639609398842e-07, |
| "loss": 0.22358329296112062, |
| "step": 990, |
| "token_acc": 0.915683082349749 |
| }, |
| { |
| "epoch": 0.007590871119214822, |
| "grad_norm": 2.2527992518345625, |
| "learning_rate": 7.590784253890754e-07, |
| "loss": 0.22756524085998536, |
| "step": 995, |
| "token_acc": 0.9161717164122998 |
| }, |
| { |
| "epoch": 0.007629016200215901, |
| "grad_norm": 2.026830062262338, |
| "learning_rate": 7.628928898382668e-07, |
| "loss": 0.22516098022460937, |
| "step": 1000, |
| "token_acc": 0.9157646132651053 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 262158, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 2, |
| "save_steps": 200, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 420470812114944.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
