| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.0, |
| "eval_steps": 500, |
| "global_step": 1300, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.007695267410542516, |
| "grad_norm": 13.97758184736854, |
| "learning_rate": 1.153846153846154e-06, |
| "loss": 1.7381, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.015390534821085032, |
| "grad_norm": 3.2654628525215252, |
| "learning_rate": 2.435897435897436e-06, |
| "loss": 1.4175, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.02308580223162755, |
| "grad_norm": 2.185550696206495, |
| "learning_rate": 3.7179487179487184e-06, |
| "loss": 1.1747, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.030781069642170065, |
| "grad_norm": 1.7452647051683785, |
| "learning_rate": 5e-06, |
| "loss": 1.0101, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.038476337052712584, |
| "grad_norm": 1.5955739897378916, |
| "learning_rate": 4.999224186876293e-06, |
| "loss": 0.9152, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.0461716044632551, |
| "grad_norm": 1.5445981337082562, |
| "learning_rate": 4.99689722901397e-06, |
| "loss": 0.8578, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.053866871873797614, |
| "grad_norm": 1.5829384919607552, |
| "learning_rate": 4.993020570640592e-06, |
| "loss": 0.819, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.06156213928434013, |
| "grad_norm": 1.5693870003526385, |
| "learning_rate": 4.987596617806111e-06, |
| "loss": 0.7848, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.06925740669488265, |
| "grad_norm": 1.444490122600457, |
| "learning_rate": 4.980628736889562e-06, |
| "loss": 0.7706, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.07695267410542517, |
| "grad_norm": 1.4885009034270107, |
| "learning_rate": 4.972121252509712e-06, |
| "loss": 0.7696, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.08464794151596768, |
| "grad_norm": 1.5903124365283623, |
| "learning_rate": 4.962079444840985e-06, |
| "loss": 0.747, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.0923432089265102, |
| "grad_norm": 1.3159285996762058, |
| "learning_rate": 4.950509546336323e-06, |
| "loss": 0.7365, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.1000384763370527, |
| "grad_norm": 1.4622841596239402, |
| "learning_rate": 4.937418737859004e-06, |
| "loss": 0.7427, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.10773374374759523, |
| "grad_norm": 1.4580401349220373, |
| "learning_rate": 4.922815144225843e-06, |
| "loss": 0.7193, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.11542901115813775, |
| "grad_norm": 1.3071477107247882, |
| "learning_rate": 4.9067078291645144e-06, |
| "loss": 0.7026, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.12312427856868026, |
| "grad_norm": 1.3745593225248909, |
| "learning_rate": 4.889106789688148e-06, |
| "loss": 0.7075, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.13081954597922277, |
| "grad_norm": 1.5581787781077656, |
| "learning_rate": 4.870022949890676e-06, |
| "loss": 0.7055, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.1385148133897653, |
| "grad_norm": 1.4683846881792455, |
| "learning_rate": 4.849468154166794e-06, |
| "loss": 0.7108, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.1462100808003078, |
| "grad_norm": 1.447276308596183, |
| "learning_rate": 4.8274551598607214e-06, |
| "loss": 0.6999, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.15390534821085033, |
| "grad_norm": 1.3637218471527626, |
| "learning_rate": 4.803997629348359e-06, |
| "loss": 0.6824, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.16160061562139286, |
| "grad_norm": 1.4688059103214917, |
| "learning_rate": 4.779110121557723e-06, |
| "loss": 0.7008, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.16929588303193535, |
| "grad_norm": 1.5334254520131108, |
| "learning_rate": 4.752808082932943e-06, |
| "loss": 0.6794, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.17699115044247787, |
| "grad_norm": 1.3954331804741134, |
| "learning_rate": 4.725107837847414e-06, |
| "loss": 0.6805, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.1846864178530204, |
| "grad_norm": 1.4979279044717357, |
| "learning_rate": 4.696026578472073e-06, |
| "loss": 0.6829, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.19238168526356292, |
| "grad_norm": 1.3658039703418836, |
| "learning_rate": 4.6655823541050575e-06, |
| "loss": 0.6821, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.2000769526741054, |
| "grad_norm": 1.395984831888077, |
| "learning_rate": 4.633794059969413e-06, |
| "loss": 0.6725, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.20777222008464794, |
| "grad_norm": 1.514025290645184, |
| "learning_rate": 4.600681425485757e-06, |
| "loss": 0.6564, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.21546748749519046, |
| "grad_norm": 1.4263355170428142, |
| "learning_rate": 4.566265002027204e-06, |
| "loss": 0.6859, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.22316275490573298, |
| "grad_norm": 1.3302752749934625, |
| "learning_rate": 4.530566150164145e-06, |
| "loss": 0.6714, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.2308580223162755, |
| "grad_norm": 1.4148249785247522, |
| "learning_rate": 4.493607026406802e-06, |
| "loss": 0.6593, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.238553289726818, |
| "grad_norm": 1.3440218448270487, |
| "learning_rate": 4.455410569453777e-06, |
| "loss": 0.6575, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.24624855713736052, |
| "grad_norm": 1.4466326720578877, |
| "learning_rate": 4.416000485955135e-06, |
| "loss": 0.6662, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.25394382454790304, |
| "grad_norm": 1.4268973112565737, |
| "learning_rate": 4.375401235798866e-06, |
| "loss": 0.6614, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.26163909195844554, |
| "grad_norm": 1.330730519119579, |
| "learning_rate": 4.333638016929835e-06, |
| "loss": 0.6539, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.2693343593689881, |
| "grad_norm": 1.397324648250839, |
| "learning_rate": 4.290736749710672e-06, |
| "loss": 0.6531, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.2770296267795306, |
| "grad_norm": 1.406291822115986, |
| "learning_rate": 4.246724060834284e-06, |
| "loss": 0.6514, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.28472489419007313, |
| "grad_norm": 1.541300053837121, |
| "learning_rate": 4.2016272667979814e-06, |
| "loss": 0.6465, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.2924201616006156, |
| "grad_norm": 1.2386876584409807, |
| "learning_rate": 4.155474356949487e-06, |
| "loss": 0.6357, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.3001154290111581, |
| "grad_norm": 1.3572343796098307, |
| "learning_rate": 4.108293976115325e-06, |
| "loss": 0.6463, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.30781069642170067, |
| "grad_norm": 1.3366293711246031, |
| "learning_rate": 4.060115406822402e-06, |
| "loss": 0.6493, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.31550596383224316, |
| "grad_norm": 1.4580177336085034, |
| "learning_rate": 4.010968551123788e-06, |
| "loss": 0.649, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.3232012312427857, |
| "grad_norm": 1.5189965677269381, |
| "learning_rate": 3.9608839120399975e-06, |
| "loss": 0.6502, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.3308964986533282, |
| "grad_norm": 1.3155120338254618, |
| "learning_rate": 3.909892574627267e-06, |
| "loss": 0.6337, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.3385917660638707, |
| "grad_norm": 1.4604046354865088, |
| "learning_rate": 3.858026186684604e-06, |
| "loss": 0.6378, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.34628703347441325, |
| "grad_norm": 1.272702075865303, |
| "learning_rate": 3.8053169391115665e-06, |
| "loss": 0.6326, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.35398230088495575, |
| "grad_norm": 1.4030248934911853, |
| "learning_rate": 3.751797545928959e-06, |
| "loss": 0.6356, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.36167756829549824, |
| "grad_norm": 1.3513670228768364, |
| "learning_rate": 3.6975012239748664e-06, |
| "loss": 0.6478, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.3693728357060408, |
| "grad_norm": 1.3948693563572099, |
| "learning_rate": 3.6424616722886004e-06, |
| "loss": 0.6286, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.3770681031165833, |
| "grad_norm": 1.3099580688712045, |
| "learning_rate": 3.586713051195378e-06, |
| "loss": 0.6211, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.38476337052712584, |
| "grad_norm": 1.4573873690579162, |
| "learning_rate": 3.530289961104698e-06, |
| "loss": 0.6284, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.39245863793766833, |
| "grad_norm": 1.343496444443226, |
| "learning_rate": 3.473227421035578e-06, |
| "loss": 0.6301, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.4001539053482108, |
| "grad_norm": 1.3989118038329986, |
| "learning_rate": 3.415560846881984e-06, |
| "loss": 0.642, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.4078491727587534, |
| "grad_norm": 1.3828624541674084, |
| "learning_rate": 3.357326029431939e-06, |
| "loss": 0.637, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.41554444016929587, |
| "grad_norm": 1.3844420582912507, |
| "learning_rate": 3.2985591121539495e-06, |
| "loss": 0.6148, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.4232397075798384, |
| "grad_norm": 1.4911796985892387, |
| "learning_rate": 3.239296568764547e-06, |
| "loss": 0.6373, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.4309349749903809, |
| "grad_norm": 1.4262609649047162, |
| "learning_rate": 3.1795751805908578e-06, |
| "loss": 0.6204, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.4386302424009234, |
| "grad_norm": 1.3783723845460145, |
| "learning_rate": 3.1194320137422483e-06, |
| "loss": 0.6253, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.44632550981146596, |
| "grad_norm": 1.2994194727386197, |
| "learning_rate": 3.0589043961052344e-06, |
| "loss": 0.629, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.45402077722200845, |
| "grad_norm": 1.4105355691796526, |
| "learning_rate": 2.9980298941759035e-06, |
| "loss": 0.6201, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.461716044632551, |
| "grad_norm": 1.2619429014746806, |
| "learning_rate": 2.936846289744252e-06, |
| "loss": 0.6167, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.4694113120430935, |
| "grad_norm": 1.3663737462582648, |
| "learning_rate": 2.875391556444898e-06, |
| "loss": 0.6236, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.477106579453636, |
| "grad_norm": 1.411619960081486, |
| "learning_rate": 2.8137038361887297e-06, |
| "loss": 0.6216, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.48480184686417854, |
| "grad_norm": 1.5379176535774535, |
| "learning_rate": 2.7518214154901025e-06, |
| "loss": 0.6066, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.49249711427472104, |
| "grad_norm": 1.408351350448584, |
| "learning_rate": 2.689782701704301e-06, |
| "loss": 0.6164, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.5001923816852636, |
| "grad_norm": 1.3308160599532777, |
| "learning_rate": 2.6276261991899915e-06, |
| "loss": 0.6169, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.5078876490958061, |
| "grad_norm": 1.2006822357761155, |
| "learning_rate": 2.5653904854114735e-06, |
| "loss": 0.604, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.5155829165063486, |
| "grad_norm": 1.326648780922408, |
| "learning_rate": 2.503114186995558e-06, |
| "loss": 0.6191, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.5232781839168911, |
| "grad_norm": 1.3684365131818692, |
| "learning_rate": 2.4408359557579294e-06, |
| "loss": 0.612, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.5309734513274337, |
| "grad_norm": 1.4972737865585115, |
| "learning_rate": 2.3785944447138804e-06, |
| "loss": 0.6112, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.5386687187379762, |
| "grad_norm": 1.3110711666835542, |
| "learning_rate": 2.316428284088296e-06, |
| "loss": 0.6135, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.5463639861485187, |
| "grad_norm": 1.4195146588151493, |
| "learning_rate": 2.254376057339789e-06, |
| "loss": 0.6051, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.5540592535590612, |
| "grad_norm": 1.194875482850615, |
| "learning_rate": 2.192476277213853e-06, |
| "loss": 0.6001, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.5617545209696037, |
| "grad_norm": 1.3178346516537736, |
| "learning_rate": 2.1307673618399078e-06, |
| "loss": 0.6075, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.5694497883801463, |
| "grad_norm": 1.3164458393737277, |
| "learning_rate": 2.0692876108870713e-06, |
| "loss": 0.6018, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.5771450557906888, |
| "grad_norm": 1.2610312676927993, |
| "learning_rate": 2.008075181793449e-06, |
| "loss": 0.6204, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.5848403232012312, |
| "grad_norm": 1.2725507486145036, |
| "learning_rate": 1.947168066083701e-06, |
| "loss": 0.6164, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.5925355906117737, |
| "grad_norm": 1.2767301793669694, |
| "learning_rate": 1.8866040657895816e-06, |
| "loss": 0.6145, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.6002308580223162, |
| "grad_norm": 1.266578085368387, |
| "learning_rate": 1.8264207699880953e-06, |
| "loss": 0.6092, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.6079261254328588, |
| "grad_norm": 1.4485986404810953, |
| "learning_rate": 1.7666555314718093e-06, |
| "loss": 0.6017, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.6156213928434013, |
| "grad_norm": 1.2764419707252663, |
| "learning_rate": 1.7073454435658294e-06, |
| "loss": 0.6139, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.6233166602539438, |
| "grad_norm": 1.390621461580982, |
| "learning_rate": 1.6485273171058074e-06, |
| "loss": 0.5992, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.6310119276644863, |
| "grad_norm": 1.3279381331045177, |
| "learning_rate": 1.5902376575912815e-06, |
| "loss": 0.6131, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.6387071950750288, |
| "grad_norm": 1.305081225242939, |
| "learning_rate": 1.5325126425285122e-06, |
| "loss": 0.6008, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.6464024624855714, |
| "grad_norm": 1.380446629763413, |
| "learning_rate": 1.4753880989769014e-06, |
| "loss": 0.6033, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.6540977298961139, |
| "grad_norm": 1.341137091405007, |
| "learning_rate": 1.4188994813129075e-06, |
| "loss": 0.5991, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.6617929973066564, |
| "grad_norm": 1.2807397862197387, |
| "learning_rate": 1.3630818492252695e-06, |
| "loss": 0.6005, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.6694882647171989, |
| "grad_norm": 1.3290585310354532, |
| "learning_rate": 1.3079698459551955e-06, |
| "loss": 0.5967, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.6771835321277414, |
| "grad_norm": 1.2054872004282364, |
| "learning_rate": 1.2535976767950111e-06, |
| "loss": 0.5908, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.6848787995382839, |
| "grad_norm": 1.3404405266973067, |
| "learning_rate": 1.1999990878586374e-06, |
| "loss": 0.5884, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.6925740669488265, |
| "grad_norm": 1.2579133452806484, |
| "learning_rate": 1.147207345137042e-06, |
| "loss": 0.594, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.700269334359369, |
| "grad_norm": 1.362754143483352, |
| "learning_rate": 1.0952552138516844e-06, |
| "loss": 0.5987, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.7079646017699115, |
| "grad_norm": 1.3973926810157402, |
| "learning_rate": 1.0441749381187718e-06, |
| "loss": 0.5868, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.715659869180454, |
| "grad_norm": 1.3683177930526904, |
| "learning_rate": 9.93998220936922e-07, |
| "loss": 0.6022, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.7233551365909965, |
| "grad_norm": 1.3162760767893575, |
| "learning_rate": 9.447562045106939e-07, |
| "loss": 0.6031, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.7310504040015391, |
| "grad_norm": 1.3999683127906501, |
| "learning_rate": 8.964794509221508e-07, |
| "loss": 0.581, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.7387456714120816, |
| "grad_norm": 1.3412546115914246, |
| "learning_rate": 8.491979231624956e-07, |
| "loss": 0.5982, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.7464409388226241, |
| "grad_norm": 1.3688820069692342, |
| "learning_rate": 8.029409665355242e-07, |
| "loss": 0.5908, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.7541362062331666, |
| "grad_norm": 1.2436627455591216, |
| "learning_rate": 7.577372904444486e-07, |
| "loss": 0.5954, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.7618314736437091, |
| "grad_norm": 1.3341305551188063, |
| "learning_rate": 7.136149505733902e-07, |
| "loss": 0.6004, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.7695267410542517, |
| "grad_norm": 1.4058866727399226, |
| "learning_rate": 6.70601331474604e-07, |
| "loss": 0.5979, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.7772220084647942, |
| "grad_norm": 1.3143787273993495, |
| "learning_rate": 6.28723129572247e-07, |
| "loss": 0.5949, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.7849172758753367, |
| "grad_norm": 1.2977338004201884, |
| "learning_rate": 5.880063365932264e-07, |
| "loss": 0.5876, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.7926125432858792, |
| "grad_norm": 1.4264492806585898, |
| "learning_rate": 5.484762234354196e-07, |
| "loss": 0.5905, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.8003078106964217, |
| "grad_norm": 1.3067082907987537, |
| "learning_rate": 5.101573244832833e-07, |
| "loss": 0.5885, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.8080030781069643, |
| "grad_norm": 1.2508997961376889, |
| "learning_rate": 4.7307342238056996e-07, |
| "loss": 0.5782, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.8156983455175068, |
| "grad_norm": 1.2843797822377727, |
| "learning_rate": 4.3724753326962634e-07, |
| "loss": 0.5941, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.8233936129280492, |
| "grad_norm": 1.3554846937801304, |
| "learning_rate": 4.0270189250640273e-07, |
| "loss": 0.5981, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.8310888803385917, |
| "grad_norm": 1.4828752168882742, |
| "learning_rate": 3.6945794086007706e-07, |
| "loss": 0.5881, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.8387841477491342, |
| "grad_norm": 1.2425957474797718, |
| "learning_rate": 3.375363112058255e-07, |
| "loss": 0.5795, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.8464794151596768, |
| "grad_norm": 1.310990751721758, |
| "learning_rate": 3.069568157190214e-07, |
| "loss": 0.59, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.8541746825702193, |
| "grad_norm": 1.3265430503182303, |
| "learning_rate": 2.777384335787969e-07, |
| "loss": 0.584, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.8618699499807618, |
| "grad_norm": 1.4117405781800352, |
| "learning_rate": 2.498992991886054e-07, |
| "loss": 0.5902, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.8695652173913043, |
| "grad_norm": 1.3047632440002366, |
| "learning_rate": 2.234566909210975e-07, |
| "loss": 0.5914, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.8772604848018468, |
| "grad_norm": 1.3474997823635917, |
| "learning_rate": 1.9842702039428923e-07, |
| "loss": 0.5892, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.8849557522123894, |
| "grad_norm": 1.1936603400315928, |
| "learning_rate": 1.7482582228568052e-07, |
| "loss": 0.5898, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.8926510196229319, |
| "grad_norm": 1.249870163608617, |
| "learning_rate": 1.5266774469065494e-07, |
| "loss": 0.5969, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.9003462870334744, |
| "grad_norm": 1.3091402765501352, |
| "learning_rate": 1.3196654003112653e-07, |
| "loss": 0.5887, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.9080415544440169, |
| "grad_norm": 1.2156507502701925, |
| "learning_rate": 1.1273505652009826e-07, |
| "loss": 0.5971, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.9157368218545594, |
| "grad_norm": 1.198249778586529, |
| "learning_rate": 9.498523018740601e-08, |
| "loss": 0.5912, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.923432089265102, |
| "grad_norm": 1.3172727281060044, |
| "learning_rate": 7.872807747162009e-08, |
| "loss": 0.5928, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.9311273566756445, |
| "grad_norm": 1.2318567742236008, |
| "learning_rate": 6.397368838268497e-08, |
| "loss": 0.5939, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.938822624086187, |
| "grad_norm": 1.275648130068706, |
| "learning_rate": 5.073122023955101e-08, |
| "loss": 0.568, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.9465178914967295, |
| "grad_norm": 1.3697978579484673, |
| "learning_rate": 3.900889198667851e-08, |
| "loss": 0.5831, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.954213158907272, |
| "grad_norm": 1.273694711826146, |
| "learning_rate": 2.8813979092947032e-08, |
| "loss": 0.5875, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.9619084263178146, |
| "grad_norm": 1.2359727905080884, |
| "learning_rate": 2.0152809036130538e-08, |
| "loss": 0.5969, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.9696036937283571, |
| "grad_norm": 1.3470864049795517, |
| "learning_rate": 1.303075737574655e-08, |
| "loss": 0.5893, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.9772989611388996, |
| "grad_norm": 1.2613200148865387, |
| "learning_rate": 7.452244416712162e-09, |
| "loss": 0.5923, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.9849942285494421, |
| "grad_norm": 1.328427554326612, |
| "learning_rate": 3.420732465878596e-09, |
| "loss": 0.5903, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.9926894959599846, |
| "grad_norm": 1.2962673063986128, |
| "learning_rate": 9.387236831495672e-10, |
| "loss": 0.5964, |
| "step": 1290 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 1.416680646437767, |
| "learning_rate": 7.758528514645313e-12, |
| "loss": 0.5932, |
| "step": 1300 |
| }, |
| { |
| "epoch": 1.0, |
| "step": 1300, |
| "total_flos": 174553276350464.0, |
| "train_loss": 0.655455028460576, |
| "train_runtime": 5388.6701, |
| "train_samples_per_second": 7.717, |
| "train_steps_per_second": 0.241 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 1300, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 174553276350464.0, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
