| { | |
| "best_metric": 0.28981047871583737, | |
| "best_model_checkpoint": "/content/drive/MyDrive/model_outputs/trial_4/checkpoint-1167", | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1167, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00856898029134533, | |
| "grad_norm": 12.96240520477295, | |
| "learning_rate": 6.233595156363815e-07, | |
| "loss": 2.1941, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01713796058269066, | |
| "grad_norm": 15.678441047668457, | |
| "learning_rate": 1.246719031272763e-06, | |
| "loss": 2.2004, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02570694087403599, | |
| "grad_norm": 19.057016372680664, | |
| "learning_rate": 1.8700785469091444e-06, | |
| "loss": 2.2096, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03427592116538132, | |
| "grad_norm": 16.71215057373047, | |
| "learning_rate": 2.493438062545526e-06, | |
| "loss": 2.2218, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04284490145672665, | |
| "grad_norm": 19.787519454956055, | |
| "learning_rate": 3.1167975781819074e-06, | |
| "loss": 2.2402, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05141388174807198, | |
| "grad_norm": 13.118367195129395, | |
| "learning_rate": 3.7401570938182888e-06, | |
| "loss": 2.1593, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05998286203941731, | |
| "grad_norm": 14.878708839416504, | |
| "learning_rate": 4.363516609454671e-06, | |
| "loss": 2.1811, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06855184233076264, | |
| "grad_norm": 26.236587524414062, | |
| "learning_rate": 4.986876125091052e-06, | |
| "loss": 2.0935, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07712082262210797, | |
| "grad_norm": 14.972529411315918, | |
| "learning_rate": 5.610235640727433e-06, | |
| "loss": 2.1434, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0856898029134533, | |
| "grad_norm": 14.05119514465332, | |
| "learning_rate": 6.233595156363815e-06, | |
| "loss": 2.0917, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.09425878320479864, | |
| "grad_norm": 13.932599067687988, | |
| "learning_rate": 6.856954672000195e-06, | |
| "loss": 2.0876, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.10282776349614396, | |
| "grad_norm": 11.802385330200195, | |
| "learning_rate": 7.4803141876365775e-06, | |
| "loss": 2.1551, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.11139674378748929, | |
| "grad_norm": 17.57957649230957, | |
| "learning_rate": 8.103673703272958e-06, | |
| "loss": 2.0485, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.11996572407883462, | |
| "grad_norm": 12.634247779846191, | |
| "learning_rate": 8.727033218909341e-06, | |
| "loss": 2.0586, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.12853470437017994, | |
| "grad_norm": 14.163841247558594, | |
| "learning_rate": 9.350392734545721e-06, | |
| "loss": 2.0832, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.13710368466152528, | |
| "grad_norm": 12.886134147644043, | |
| "learning_rate": 9.973752250182104e-06, | |
| "loss": 2.0785, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.1456726649528706, | |
| "grad_norm": 12.337018966674805, | |
| "learning_rate": 1.0597111765818484e-05, | |
| "loss": 2.1782, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.15424164524421594, | |
| "grad_norm": 15.206589698791504, | |
| "learning_rate": 1.1220471281454867e-05, | |
| "loss": 2.05, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.16281062553556128, | |
| "grad_norm": 15.637868881225586, | |
| "learning_rate": 1.1843830797091246e-05, | |
| "loss": 2.0865, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1713796058269066, | |
| "grad_norm": 13.318354606628418, | |
| "learning_rate": 1.246719031272763e-05, | |
| "loss": 1.9896, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.17994858611825193, | |
| "grad_norm": 15.895234107971191, | |
| "learning_rate": 1.3090549828364011e-05, | |
| "loss": 2.2041, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.18851756640959727, | |
| "grad_norm": 11.681656837463379, | |
| "learning_rate": 1.3580958892325523e-05, | |
| "loss": 2.0332, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.19708654670094258, | |
| "grad_norm": 12.1074800491333, | |
| "learning_rate": 1.353956614958756e-05, | |
| "loss": 1.9992, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.20565552699228792, | |
| "grad_norm": 16.131893157958984, | |
| "learning_rate": 1.3498173406849598e-05, | |
| "loss": 2.0392, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.21422450728363324, | |
| "grad_norm": 13.486928939819336, | |
| "learning_rate": 1.3456780664111635e-05, | |
| "loss": 2.0591, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.22279348757497858, | |
| "grad_norm": 12.962348937988281, | |
| "learning_rate": 1.3415387921373673e-05, | |
| "loss": 2.1737, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.23136246786632392, | |
| "grad_norm": 15.706825256347656, | |
| "learning_rate": 1.337399517863571e-05, | |
| "loss": 2.0788, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.23993144815766923, | |
| "grad_norm": 15.435516357421875, | |
| "learning_rate": 1.3332602435897748e-05, | |
| "loss": 2.1001, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.24850042844901457, | |
| "grad_norm": 10.667174339294434, | |
| "learning_rate": 1.3291209693159784e-05, | |
| "loss": 1.9447, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.2570694087403599, | |
| "grad_norm": 14.559552192687988, | |
| "learning_rate": 1.3249816950421821e-05, | |
| "loss": 2.0502, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2656383890317052, | |
| "grad_norm": 15.85730266571045, | |
| "learning_rate": 1.320842420768386e-05, | |
| "loss": 2.0924, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.27420736932305056, | |
| "grad_norm": 13.333989143371582, | |
| "learning_rate": 1.3167031464945898e-05, | |
| "loss": 2.1221, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.2827763496143959, | |
| "grad_norm": 13.683819770812988, | |
| "learning_rate": 1.3125638722207935e-05, | |
| "loss": 2.1562, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2913453299057412, | |
| "grad_norm": 12.066128730773926, | |
| "learning_rate": 1.3084245979469973e-05, | |
| "loss": 2.0211, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.29991431019708653, | |
| "grad_norm": 14.573673248291016, | |
| "learning_rate": 1.3042853236732009e-05, | |
| "loss": 2.0365, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.30848329048843187, | |
| "grad_norm": 19.47612762451172, | |
| "learning_rate": 1.3001460493994046e-05, | |
| "loss": 1.9834, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.3170522707797772, | |
| "grad_norm": 13.694817543029785, | |
| "learning_rate": 1.2960067751256084e-05, | |
| "loss": 2.0116, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.32562125107112255, | |
| "grad_norm": 12.057513236999512, | |
| "learning_rate": 1.2918675008518121e-05, | |
| "loss": 2.0626, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.3341902313624679, | |
| "grad_norm": 12.8948974609375, | |
| "learning_rate": 1.2877282265780159e-05, | |
| "loss": 1.9787, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3427592116538132, | |
| "grad_norm": 14.375242233276367, | |
| "learning_rate": 1.2835889523042196e-05, | |
| "loss": 2.0237, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3513281919451585, | |
| "grad_norm": 13.976824760437012, | |
| "learning_rate": 1.2794496780304234e-05, | |
| "loss": 2.1159, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.35989717223650386, | |
| "grad_norm": 15.234496116638184, | |
| "learning_rate": 1.2753104037566271e-05, | |
| "loss": 1.9116, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.3684661525278492, | |
| "grad_norm": 12.984347343444824, | |
| "learning_rate": 1.2711711294828309e-05, | |
| "loss": 2.0484, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.37703513281919454, | |
| "grad_norm": 10.440543174743652, | |
| "learning_rate": 1.2670318552090346e-05, | |
| "loss": 2.0348, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3856041131105398, | |
| "grad_norm": 11.408565521240234, | |
| "learning_rate": 1.2628925809352384e-05, | |
| "loss": 2.0866, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.39417309340188517, | |
| "grad_norm": 15.430132865905762, | |
| "learning_rate": 1.2587533066614421e-05, | |
| "loss": 2.0252, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.4027420736932305, | |
| "grad_norm": 15.553166389465332, | |
| "learning_rate": 1.2546140323876457e-05, | |
| "loss": 2.0838, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.41131105398457585, | |
| "grad_norm": 14.190583229064941, | |
| "learning_rate": 1.2504747581138496e-05, | |
| "loss": 2.0674, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.4198800342759212, | |
| "grad_norm": 19.912630081176758, | |
| "learning_rate": 1.2463354838400534e-05, | |
| "loss": 2.0175, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.4284490145672665, | |
| "grad_norm": 11.050692558288574, | |
| "learning_rate": 1.2421962095662571e-05, | |
| "loss": 2.0227, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4370179948586118, | |
| "grad_norm": 13.648895263671875, | |
| "learning_rate": 1.2380569352924609e-05, | |
| "loss": 2.1153, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.44558697514995715, | |
| "grad_norm": 13.396595001220703, | |
| "learning_rate": 1.2339176610186645e-05, | |
| "loss": 2.0524, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.4541559554413025, | |
| "grad_norm": 12.05173110961914, | |
| "learning_rate": 1.2297783867448682e-05, | |
| "loss": 2.0801, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.46272493573264784, | |
| "grad_norm": 11.418408393859863, | |
| "learning_rate": 1.225639112471072e-05, | |
| "loss": 2.0219, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4712939160239931, | |
| "grad_norm": 14.92882251739502, | |
| "learning_rate": 1.2214998381972757e-05, | |
| "loss": 1.9197, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.47986289631533846, | |
| "grad_norm": 13.67534351348877, | |
| "learning_rate": 1.2173605639234796e-05, | |
| "loss": 1.9579, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4884318766066838, | |
| "grad_norm": 16.3277645111084, | |
| "learning_rate": 1.2132212896496834e-05, | |
| "loss": 2.0499, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.49700085689802914, | |
| "grad_norm": 12.686991691589355, | |
| "learning_rate": 1.209082015375887e-05, | |
| "loss": 1.8892, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.5055698371893744, | |
| "grad_norm": 14.13610553741455, | |
| "learning_rate": 1.2049427411020907e-05, | |
| "loss": 1.9821, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.5141388174807198, | |
| "grad_norm": 10.20384693145752, | |
| "learning_rate": 1.2008034668282945e-05, | |
| "loss": 1.8765, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.5227077977720651, | |
| "grad_norm": 11.100608825683594, | |
| "learning_rate": 1.1966641925544982e-05, | |
| "loss": 1.928, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.5312767780634104, | |
| "grad_norm": 13.737257957458496, | |
| "learning_rate": 1.192524918280702e-05, | |
| "loss": 2.0266, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.5398457583547558, | |
| "grad_norm": 13.313102722167969, | |
| "learning_rate": 1.1883856440069057e-05, | |
| "loss": 1.9792, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5484147386461011, | |
| "grad_norm": 16.294010162353516, | |
| "learning_rate": 1.1842463697331093e-05, | |
| "loss": 2.0274, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5569837189374465, | |
| "grad_norm": 14.80037784576416, | |
| "learning_rate": 1.1801070954593132e-05, | |
| "loss": 2.0204, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5655526992287918, | |
| "grad_norm": 16.782167434692383, | |
| "learning_rate": 1.175967821185517e-05, | |
| "loss": 1.896, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5741216795201372, | |
| "grad_norm": 14.986900329589844, | |
| "learning_rate": 1.1718285469117207e-05, | |
| "loss": 1.9126, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.5826906598114824, | |
| "grad_norm": 15.64176082611084, | |
| "learning_rate": 1.1676892726379245e-05, | |
| "loss": 1.8983, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5912596401028277, | |
| "grad_norm": 14.483860969543457, | |
| "learning_rate": 1.1635499983641282e-05, | |
| "loss": 2.1211, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5998286203941731, | |
| "grad_norm": 9.888971328735352, | |
| "learning_rate": 1.1594107240903318e-05, | |
| "loss": 1.8572, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.6083976006855184, | |
| "grad_norm": 12.11154556274414, | |
| "learning_rate": 1.1552714498165355e-05, | |
| "loss": 1.866, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.6169665809768637, | |
| "grad_norm": 14.539010047912598, | |
| "learning_rate": 1.1511321755427393e-05, | |
| "loss": 1.9029, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.6255355612682091, | |
| "grad_norm": 17.459091186523438, | |
| "learning_rate": 1.1469929012689432e-05, | |
| "loss": 1.9333, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.6341045415595544, | |
| "grad_norm": 14.461770057678223, | |
| "learning_rate": 1.142853626995147e-05, | |
| "loss": 2.0981, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.6426735218508998, | |
| "grad_norm": 15.937264442443848, | |
| "learning_rate": 1.1387143527213507e-05, | |
| "loss": 1.8607, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.6512425021422451, | |
| "grad_norm": 16.7382869720459, | |
| "learning_rate": 1.1345750784475543e-05, | |
| "loss": 2.1263, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6598114824335904, | |
| "grad_norm": 15.768759727478027, | |
| "learning_rate": 1.130435804173758e-05, | |
| "loss": 1.928, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6683804627249358, | |
| "grad_norm": 18.80556869506836, | |
| "learning_rate": 1.1262965298999618e-05, | |
| "loss": 1.98, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.676949443016281, | |
| "grad_norm": 14.333468437194824, | |
| "learning_rate": 1.1221572556261655e-05, | |
| "loss": 1.9067, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6855184233076264, | |
| "grad_norm": 19.20866584777832, | |
| "learning_rate": 1.1180179813523693e-05, | |
| "loss": 2.0511, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6940874035989717, | |
| "grad_norm": 18.691129684448242, | |
| "learning_rate": 1.113878707078573e-05, | |
| "loss": 1.872, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.702656383890317, | |
| "grad_norm": 14.753878593444824, | |
| "learning_rate": 1.1097394328047768e-05, | |
| "loss": 1.8993, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.7112253641816624, | |
| "grad_norm": 13.987009048461914, | |
| "learning_rate": 1.1056001585309806e-05, | |
| "loss": 1.8868, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.7197943444730077, | |
| "grad_norm": 16.61811637878418, | |
| "learning_rate": 1.1014608842571843e-05, | |
| "loss": 1.8531, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.7283633247643531, | |
| "grad_norm": 18.929941177368164, | |
| "learning_rate": 1.097321609983388e-05, | |
| "loss": 1.8008, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.7369323050556984, | |
| "grad_norm": 18.89458465576172, | |
| "learning_rate": 1.0931823357095918e-05, | |
| "loss": 1.935, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.7455012853470437, | |
| "grad_norm": 15.524568557739258, | |
| "learning_rate": 1.0890430614357956e-05, | |
| "loss": 2.0548, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.7540702656383891, | |
| "grad_norm": 17.038110733032227, | |
| "learning_rate": 1.0849037871619991e-05, | |
| "loss": 2.0176, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.7626392459297343, | |
| "grad_norm": 16.24259376525879, | |
| "learning_rate": 1.0807645128882029e-05, | |
| "loss": 2.0311, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.7712082262210797, | |
| "grad_norm": 12.702564239501953, | |
| "learning_rate": 1.0766252386144068e-05, | |
| "loss": 1.9613, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.779777206512425, | |
| "grad_norm": 13.47549057006836, | |
| "learning_rate": 1.0724859643406106e-05, | |
| "loss": 1.9554, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7883461868037703, | |
| "grad_norm": 15.315031051635742, | |
| "learning_rate": 1.0683466900668143e-05, | |
| "loss": 1.8628, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7969151670951157, | |
| "grad_norm": 12.436241149902344, | |
| "learning_rate": 1.0642074157930179e-05, | |
| "loss": 1.9767, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.805484147386461, | |
| "grad_norm": 17.100671768188477, | |
| "learning_rate": 1.0600681415192216e-05, | |
| "loss": 2.0268, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.8140531276778064, | |
| "grad_norm": 14.803923606872559, | |
| "learning_rate": 1.0559288672454254e-05, | |
| "loss": 2.0516, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.8226221079691517, | |
| "grad_norm": 17.308460235595703, | |
| "learning_rate": 1.0517895929716291e-05, | |
| "loss": 1.9354, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.831191088260497, | |
| "grad_norm": 38.664039611816406, | |
| "learning_rate": 1.0476503186978329e-05, | |
| "loss": 1.9623, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.8397600685518424, | |
| "grad_norm": 16.550312042236328, | |
| "learning_rate": 1.0435110444240368e-05, | |
| "loss": 1.888, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.8483290488431876, | |
| "grad_norm": 19.344846725463867, | |
| "learning_rate": 1.0393717701502404e-05, | |
| "loss": 1.9992, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.856898029134533, | |
| "grad_norm": 18.766752243041992, | |
| "learning_rate": 1.0352324958764441e-05, | |
| "loss": 1.8824, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.8654670094258783, | |
| "grad_norm": 20.725662231445312, | |
| "learning_rate": 1.0310932216026479e-05, | |
| "loss": 1.7351, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.8740359897172236, | |
| "grad_norm": 15.772370338439941, | |
| "learning_rate": 1.0269539473288516e-05, | |
| "loss": 1.9839, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.882604970008569, | |
| "grad_norm": 15.904685974121094, | |
| "learning_rate": 1.0228146730550554e-05, | |
| "loss": 2.1096, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.8911739502999143, | |
| "grad_norm": 19.53571891784668, | |
| "learning_rate": 1.0186753987812591e-05, | |
| "loss": 1.9122, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8997429305912596, | |
| "grad_norm": 14.403525352478027, | |
| "learning_rate": 1.0145361245074627e-05, | |
| "loss": 1.8495, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.908311910882605, | |
| "grad_norm": 20.06512451171875, | |
| "learning_rate": 1.0103968502336666e-05, | |
| "loss": 1.9467, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.9168808911739503, | |
| "grad_norm": 15.332779884338379, | |
| "learning_rate": 1.0062575759598704e-05, | |
| "loss": 1.7918, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.9254498714652957, | |
| "grad_norm": 18.091779708862305, | |
| "learning_rate": 1.0021183016860741e-05, | |
| "loss": 1.8323, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.934018851756641, | |
| "grad_norm": 17.92203712463379, | |
| "learning_rate": 9.979790274122779e-06, | |
| "loss": 1.9448, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.9425878320479862, | |
| "grad_norm": 11.862146377563477, | |
| "learning_rate": 9.938397531384816e-06, | |
| "loss": 1.9453, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.9511568123393316, | |
| "grad_norm": 16.67616844177246, | |
| "learning_rate": 9.897004788646852e-06, | |
| "loss": 1.9977, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.9597257926306769, | |
| "grad_norm": 17.18949317932129, | |
| "learning_rate": 9.85561204590889e-06, | |
| "loss": 1.8888, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.9682947729220223, | |
| "grad_norm": 19.521203994750977, | |
| "learning_rate": 9.814219303170927e-06, | |
| "loss": 1.948, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.9768637532133676, | |
| "grad_norm": 21.371353149414062, | |
| "learning_rate": 9.772826560432965e-06, | |
| "loss": 1.9285, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.9854327335047129, | |
| "grad_norm": 18.078819274902344, | |
| "learning_rate": 9.731433817695004e-06, | |
| "loss": 1.9144, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9940017137960583, | |
| "grad_norm": 26.718231201171875, | |
| "learning_rate": 9.690041074957041e-06, | |
| "loss": 1.8113, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_classification_report": { | |
| "accuracy": 0.3035, | |
| "ar": { | |
| "f1-score": 0.23214285714285715, | |
| "precision": 0.3, | |
| "recall": 0.18932038834951456, | |
| "support": 206.0 | |
| }, | |
| "cl": { | |
| "f1-score": 0.2225609756097561, | |
| "precision": 0.1994535519125683, | |
| "recall": 0.2517241379310345, | |
| "support": 290.0 | |
| }, | |
| "co": { | |
| "f1-score": 0.3453038674033149, | |
| "precision": 0.28868360277136257, | |
| "recall": 0.42955326460481097, | |
| "support": 291.0 | |
| }, | |
| "es": { | |
| "f1-score": 0.3566666666666667, | |
| "precision": 0.3333333333333333, | |
| "recall": 0.3835125448028674, | |
| "support": 279.0 | |
| }, | |
| "macro avg": { | |
| "f1-score": 0.28981047871583737, | |
| "precision": 0.31679889037420644, | |
| "recall": 0.28794417545744694, | |
| "support": 2000.0 | |
| }, | |
| "mx": { | |
| "f1-score": 0.2676767676767677, | |
| "precision": 0.5047619047619047, | |
| "recall": 0.18213058419243985, | |
| "support": 291.0 | |
| }, | |
| "pe": { | |
| "f1-score": 0.27666666666666667, | |
| "precision": 0.2686084142394822, | |
| "recall": 0.2852233676975945, | |
| "support": 291.0 | |
| }, | |
| "pr": { | |
| "f1-score": 0.6162162162162163, | |
| "precision": 0.6785714285714286, | |
| "recall": 0.5643564356435643, | |
| "support": 101.0 | |
| }, | |
| "uy": { | |
| "f1-score": 0.2910602910602911, | |
| "precision": 0.2777777777777778, | |
| "recall": 0.3056768558951965, | |
| "support": 229.0 | |
| }, | |
| "ve": { | |
| "f1-score": 0.0, | |
| "precision": 0.0, | |
| "recall": 0.0, | |
| "support": 22.0 | |
| }, | |
| "weighted avg": { | |
| "f1-score": 0.2998260603986032, | |
| "precision": 0.3269230233436701, | |
| "recall": 0.3035, | |
| "support": 2000.0 | |
| } | |
| }, | |
| "eval_f1": 0.28981047871583737, | |
| "eval_loss": 1.8416967391967773, | |
| "eval_runtime": 5.6259, | |
| "eval_samples_per_second": 355.502, | |
| "eval_steps_per_second": 88.875, | |
| "step": 1167 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 3501, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 153469167996672.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |