| { | |
| "best_metric": 0.961, | |
| "best_model_checkpoint": "clip-vit-base-patch32-finetuned-eurosat/checkpoint-1053", | |
| "epoch": 2.9936034115138592, | |
| "eval_steps": 500, | |
| "global_step": 1053, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.028429282160625444, | |
| "grad_norm": 12.128193855285645, | |
| "learning_rate": 4.716981132075472e-06, | |
| "loss": 2.3369, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05685856432125089, | |
| "grad_norm": 19.345117568969727, | |
| "learning_rate": 9.433962264150944e-06, | |
| "loss": 2.0309, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08528784648187633, | |
| "grad_norm": 39.82497787475586, | |
| "learning_rate": 1.4150943396226415e-05, | |
| "loss": 1.466, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.11371712864250177, | |
| "grad_norm": 39.60441589355469, | |
| "learning_rate": 1.8867924528301888e-05, | |
| "loss": 1.2444, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.14214641080312723, | |
| "grad_norm": 40.64250946044922, | |
| "learning_rate": 2.358490566037736e-05, | |
| "loss": 1.0176, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17057569296375266, | |
| "grad_norm": 39.66933822631836, | |
| "learning_rate": 2.830188679245283e-05, | |
| "loss": 0.9046, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.19900497512437812, | |
| "grad_norm": 30.824308395385742, | |
| "learning_rate": 3.30188679245283e-05, | |
| "loss": 0.8737, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.22743425728500355, | |
| "grad_norm": 26.433198928833008, | |
| "learning_rate": 3.7735849056603776e-05, | |
| "loss": 0.8393, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.255863539445629, | |
| "grad_norm": 21.772321701049805, | |
| "learning_rate": 4.245283018867925e-05, | |
| "loss": 0.8792, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.28429282160625446, | |
| "grad_norm": 23.111845016479492, | |
| "learning_rate": 4.716981132075472e-05, | |
| "loss": 0.9117, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.31272210376687987, | |
| "grad_norm": 18.194908142089844, | |
| "learning_rate": 4.978880675818374e-05, | |
| "loss": 0.7778, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3411513859275053, | |
| "grad_norm": 20.22426986694336, | |
| "learning_rate": 4.9260823653643085e-05, | |
| "loss": 0.8241, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3695806680881308, | |
| "grad_norm": 21.78709602355957, | |
| "learning_rate": 4.8732840549102435e-05, | |
| "loss": 0.7939, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.39800995024875624, | |
| "grad_norm": 19.71605682373047, | |
| "learning_rate": 4.820485744456177e-05, | |
| "loss": 0.876, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.42643923240938164, | |
| "grad_norm": 18.039392471313477, | |
| "learning_rate": 4.767687434002112e-05, | |
| "loss": 0.768, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4548685145700071, | |
| "grad_norm": 19.62349510192871, | |
| "learning_rate": 4.7148891235480466e-05, | |
| "loss": 0.7447, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.48329779673063256, | |
| "grad_norm": 15.236215591430664, | |
| "learning_rate": 4.662090813093981e-05, | |
| "loss": 0.8356, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.511727078891258, | |
| "grad_norm": 19.196414947509766, | |
| "learning_rate": 4.609292502639916e-05, | |
| "loss": 0.8388, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5401563610518835, | |
| "grad_norm": 23.391098022460938, | |
| "learning_rate": 4.55649419218585e-05, | |
| "loss": 0.8099, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5685856432125089, | |
| "grad_norm": 14.287452697753906, | |
| "learning_rate": 4.503695881731785e-05, | |
| "loss": 0.7713, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5970149253731343, | |
| "grad_norm": 21.557052612304688, | |
| "learning_rate": 4.45089757127772e-05, | |
| "loss": 0.7324, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.6254442075337597, | |
| "grad_norm": 18.124984741210938, | |
| "learning_rate": 4.398099260823654e-05, | |
| "loss": 0.7383, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6538734896943852, | |
| "grad_norm": 20.559268951416016, | |
| "learning_rate": 4.3453009503695884e-05, | |
| "loss": 0.728, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6823027718550106, | |
| "grad_norm": 16.50807762145996, | |
| "learning_rate": 4.292502639915523e-05, | |
| "loss": 0.7594, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7107320540156361, | |
| "grad_norm": 18.29205894470215, | |
| "learning_rate": 4.239704329461457e-05, | |
| "loss": 0.7597, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.7391613361762616, | |
| "grad_norm": 17.540321350097656, | |
| "learning_rate": 4.186906019007392e-05, | |
| "loss": 0.6979, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.767590618336887, | |
| "grad_norm": 18.201658248901367, | |
| "learning_rate": 4.1341077085533265e-05, | |
| "loss": 0.7461, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.7960199004975125, | |
| "grad_norm": 17.44607162475586, | |
| "learning_rate": 4.081309398099261e-05, | |
| "loss": 0.6786, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.8244491826581379, | |
| "grad_norm": 14.286916732788086, | |
| "learning_rate": 4.028511087645195e-05, | |
| "loss": 0.6731, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.8528784648187633, | |
| "grad_norm": 13.381218910217285, | |
| "learning_rate": 3.97571277719113e-05, | |
| "loss": 0.6752, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8813077469793887, | |
| "grad_norm": 11.218981742858887, | |
| "learning_rate": 3.9229144667370646e-05, | |
| "loss": 0.6389, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.9097370291400142, | |
| "grad_norm": 11.469705581665039, | |
| "learning_rate": 3.870116156282999e-05, | |
| "loss": 0.643, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.9381663113006397, | |
| "grad_norm": 17.61397933959961, | |
| "learning_rate": 3.817317845828934e-05, | |
| "loss": 0.6538, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.9665955934612651, | |
| "grad_norm": 15.262922286987305, | |
| "learning_rate": 3.764519535374868e-05, | |
| "loss": 0.6231, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.9950248756218906, | |
| "grad_norm": 15.590916633605957, | |
| "learning_rate": 3.711721224920803e-05, | |
| "loss": 0.6556, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.997867803837953, | |
| "eval_accuracy": 0.9068, | |
| "eval_loss": 0.28233855962753296, | |
| "eval_runtime": 15.6893, | |
| "eval_samples_per_second": 318.688, | |
| "eval_steps_per_second": 10.007, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.023454157782516, | |
| "grad_norm": 14.132500648498535, | |
| "learning_rate": 3.658922914466738e-05, | |
| "loss": 0.5939, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.0518834399431414, | |
| "grad_norm": 14.34232234954834, | |
| "learning_rate": 3.6061246040126714e-05, | |
| "loss": 0.5661, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.080312722103767, | |
| "grad_norm": 23.517648696899414, | |
| "learning_rate": 3.5533262935586064e-05, | |
| "loss": 0.6292, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.1087420042643923, | |
| "grad_norm": 10.964395523071289, | |
| "learning_rate": 3.500527983104541e-05, | |
| "loss": 0.5544, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.1371712864250179, | |
| "grad_norm": 12.745725631713867, | |
| "learning_rate": 3.447729672650475e-05, | |
| "loss": 0.5386, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.1656005685856432, | |
| "grad_norm": 13.943474769592285, | |
| "learning_rate": 3.3949313621964095e-05, | |
| "loss": 0.5936, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.1940298507462686, | |
| "grad_norm": 16.283620834350586, | |
| "learning_rate": 3.3421330517423445e-05, | |
| "loss": 0.5046, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.2224591329068941, | |
| "grad_norm": 12.565506935119629, | |
| "learning_rate": 3.289334741288279e-05, | |
| "loss": 0.578, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.2508884150675195, | |
| "grad_norm": 12.695046424865723, | |
| "learning_rate": 3.236536430834213e-05, | |
| "loss": 0.5875, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.279317697228145, | |
| "grad_norm": 15.638214111328125, | |
| "learning_rate": 3.183738120380148e-05, | |
| "loss": 0.5058, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.3077469793887704, | |
| "grad_norm": 13.463284492492676, | |
| "learning_rate": 3.130939809926082e-05, | |
| "loss": 0.5279, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.336176261549396, | |
| "grad_norm": 10.782665252685547, | |
| "learning_rate": 3.078141499472017e-05, | |
| "loss": 0.542, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.3646055437100213, | |
| "grad_norm": 18.311925888061523, | |
| "learning_rate": 3.0253431890179517e-05, | |
| "loss": 0.5147, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.3930348258706466, | |
| "grad_norm": 12.827178955078125, | |
| "learning_rate": 2.972544878563886e-05, | |
| "loss": 0.5325, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.4214641080312722, | |
| "grad_norm": 13.880022048950195, | |
| "learning_rate": 2.9197465681098207e-05, | |
| "loss": 0.5294, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.4498933901918978, | |
| "grad_norm": 15.585273742675781, | |
| "learning_rate": 2.8669482576557548e-05, | |
| "loss": 0.532, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.4783226723525231, | |
| "grad_norm": 12.55144214630127, | |
| "learning_rate": 2.8141499472016898e-05, | |
| "loss": 0.52, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.5067519545131485, | |
| "grad_norm": 13.474200248718262, | |
| "learning_rate": 2.7613516367476245e-05, | |
| "loss": 0.5257, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.535181236673774, | |
| "grad_norm": 12.359495162963867, | |
| "learning_rate": 2.7085533262935585e-05, | |
| "loss": 0.5541, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.5636105188343994, | |
| "grad_norm": 12.825304985046387, | |
| "learning_rate": 2.6557550158394935e-05, | |
| "loss": 0.4987, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.5920398009950247, | |
| "grad_norm": 10.33250904083252, | |
| "learning_rate": 2.6029567053854276e-05, | |
| "loss": 0.4908, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.6204690831556503, | |
| "grad_norm": 13.17313289642334, | |
| "learning_rate": 2.5501583949313622e-05, | |
| "loss": 0.4884, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.6488983653162759, | |
| "grad_norm": 12.920696258544922, | |
| "learning_rate": 2.497360084477297e-05, | |
| "loss": 0.4873, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.6773276474769012, | |
| "grad_norm": 11.354279518127441, | |
| "learning_rate": 2.4445617740232313e-05, | |
| "loss": 0.4974, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.7057569296375266, | |
| "grad_norm": 13.763202667236328, | |
| "learning_rate": 2.391763463569166e-05, | |
| "loss": 0.4398, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.7341862117981521, | |
| "grad_norm": 8.816915512084961, | |
| "learning_rate": 2.3389651531151003e-05, | |
| "loss": 0.4761, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.7626154939587777, | |
| "grad_norm": 13.627618789672852, | |
| "learning_rate": 2.286166842661035e-05, | |
| "loss": 0.469, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.7910447761194028, | |
| "grad_norm": 17.703720092773438, | |
| "learning_rate": 2.2333685322069694e-05, | |
| "loss": 0.4148, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.8194740582800284, | |
| "grad_norm": 9.903636932373047, | |
| "learning_rate": 2.180570221752904e-05, | |
| "loss": 0.4824, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.847903340440654, | |
| "grad_norm": 10.972967147827148, | |
| "learning_rate": 2.1277719112988384e-05, | |
| "loss": 0.4282, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.8763326226012793, | |
| "grad_norm": 12.979598045349121, | |
| "learning_rate": 2.074973600844773e-05, | |
| "loss": 0.5252, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.9047619047619047, | |
| "grad_norm": 11.675236701965332, | |
| "learning_rate": 2.0221752903907075e-05, | |
| "loss": 0.4225, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.9331911869225302, | |
| "grad_norm": 13.110438346862793, | |
| "learning_rate": 1.9693769799366422e-05, | |
| "loss": 0.4313, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.9616204690831558, | |
| "grad_norm": 8.968453407287598, | |
| "learning_rate": 1.9165786694825765e-05, | |
| "loss": 0.4725, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.9900497512437811, | |
| "grad_norm": 13.918693542480469, | |
| "learning_rate": 1.863780359028511e-05, | |
| "loss": 0.4659, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.9985785358919688, | |
| "eval_accuracy": 0.9474, | |
| "eval_loss": 0.15807794034481049, | |
| "eval_runtime": 15.4621, | |
| "eval_samples_per_second": 323.371, | |
| "eval_steps_per_second": 10.154, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 2.0184790334044065, | |
| "grad_norm": 10.97941780090332, | |
| "learning_rate": 1.810982048574446e-05, | |
| "loss": 0.4331, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.046908315565032, | |
| "grad_norm": 12.966859817504883, | |
| "learning_rate": 1.7581837381203803e-05, | |
| "loss": 0.4168, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.0753375977256576, | |
| "grad_norm": 13.996347427368164, | |
| "learning_rate": 1.7053854276663146e-05, | |
| "loss": 0.3803, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.1037668798862827, | |
| "grad_norm": 11.576925277709961, | |
| "learning_rate": 1.6525871172122493e-05, | |
| "loss": 0.3876, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.1321961620469083, | |
| "grad_norm": 13.173456192016602, | |
| "learning_rate": 1.5997888067581837e-05, | |
| "loss": 0.3753, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.160625444207534, | |
| "grad_norm": 10.2720365524292, | |
| "learning_rate": 1.5469904963041184e-05, | |
| "loss": 0.3442, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.189054726368159, | |
| "grad_norm": 12.404537200927734, | |
| "learning_rate": 1.4941921858500529e-05, | |
| "loss": 0.3509, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.2174840085287846, | |
| "grad_norm": 12.196104049682617, | |
| "learning_rate": 1.4413938753959874e-05, | |
| "loss": 0.3653, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.24591329068941, | |
| "grad_norm": 10.60234546661377, | |
| "learning_rate": 1.388595564941922e-05, | |
| "loss": 0.4059, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.2743425728500357, | |
| "grad_norm": 11.045681953430176, | |
| "learning_rate": 1.3357972544878563e-05, | |
| "loss": 0.3827, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.302771855010661, | |
| "grad_norm": 11.341653823852539, | |
| "learning_rate": 1.2829989440337912e-05, | |
| "loss": 0.3632, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.3312011371712864, | |
| "grad_norm": 9.08681869506836, | |
| "learning_rate": 1.2302006335797255e-05, | |
| "loss": 0.3687, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.359630419331912, | |
| "grad_norm": 11.295198440551758, | |
| "learning_rate": 1.17740232312566e-05, | |
| "loss": 0.3599, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.388059701492537, | |
| "grad_norm": 9.960636138916016, | |
| "learning_rate": 1.1246040126715946e-05, | |
| "loss": 0.3642, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.4164889836531627, | |
| "grad_norm": 12.375933647155762, | |
| "learning_rate": 1.0718057022175291e-05, | |
| "loss": 0.3472, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.4449182658137882, | |
| "grad_norm": 10.313027381896973, | |
| "learning_rate": 1.0190073917634636e-05, | |
| "loss": 0.3668, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.473347547974414, | |
| "grad_norm": 11.329744338989258, | |
| "learning_rate": 9.662090813093982e-06, | |
| "loss": 0.3144, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.501776830135039, | |
| "grad_norm": 11.012008666992188, | |
| "learning_rate": 9.134107708553327e-06, | |
| "loss": 0.3482, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.5302061122956645, | |
| "grad_norm": 8.074384689331055, | |
| "learning_rate": 8.606124604012672e-06, | |
| "loss": 0.3472, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.55863539445629, | |
| "grad_norm": 11.275156021118164, | |
| "learning_rate": 8.078141499472017e-06, | |
| "loss": 0.3527, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.587064676616915, | |
| "grad_norm": 11.760042190551758, | |
| "learning_rate": 7.5501583949313625e-06, | |
| "loss": 0.3503, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.6154939587775408, | |
| "grad_norm": 9.786343574523926, | |
| "learning_rate": 7.022175290390708e-06, | |
| "loss": 0.3338, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.6439232409381663, | |
| "grad_norm": 10.899317741394043, | |
| "learning_rate": 6.494192185850054e-06, | |
| "loss": 0.3648, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.672352523098792, | |
| "grad_norm": 9.24864673614502, | |
| "learning_rate": 5.966209081309398e-06, | |
| "loss": 0.3346, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.7007818052594175, | |
| "grad_norm": 10.97048282623291, | |
| "learning_rate": 5.438225976768744e-06, | |
| "loss": 0.338, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.7292110874200426, | |
| "grad_norm": 9.632309913635254, | |
| "learning_rate": 4.910242872228089e-06, | |
| "loss": 0.2918, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.757640369580668, | |
| "grad_norm": 9.58181381225586, | |
| "learning_rate": 4.382259767687434e-06, | |
| "loss": 0.3197, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.7860696517412933, | |
| "grad_norm": 11.46987247467041, | |
| "learning_rate": 3.854276663146779e-06, | |
| "loss": 0.2981, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.814498933901919, | |
| "grad_norm": 14.371786117553711, | |
| "learning_rate": 3.326293558606125e-06, | |
| "loss": 0.327, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.8429282160625444, | |
| "grad_norm": 11.488155364990234, | |
| "learning_rate": 2.79831045406547e-06, | |
| "loss": 0.3284, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.87135749822317, | |
| "grad_norm": 13.239998817443848, | |
| "learning_rate": 2.2703273495248154e-06, | |
| "loss": 0.3365, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.8997867803837956, | |
| "grad_norm": 10.396666526794434, | |
| "learning_rate": 1.7423442449841606e-06, | |
| "loss": 0.3238, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.9282160625444207, | |
| "grad_norm": 12.260651588439941, | |
| "learning_rate": 1.2143611404435059e-06, | |
| "loss": 0.3026, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.9566453447050463, | |
| "grad_norm": 9.99549674987793, | |
| "learning_rate": 6.863780359028511e-07, | |
| "loss": 0.2994, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.9850746268656714, | |
| "grad_norm": 9.32156753540039, | |
| "learning_rate": 1.5839493136219642e-07, | |
| "loss": 0.2639, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.9936034115138592, | |
| "eval_accuracy": 0.961, | |
| "eval_loss": 0.11085611581802368, | |
| "eval_runtime": 15.421, | |
| "eval_samples_per_second": 324.234, | |
| "eval_steps_per_second": 10.181, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 2.9936034115138592, | |
| "step": 1053, | |
| "total_flos": 1.0638717415807058e+19, | |
| "train_loss": 0.580430903835514, | |
| "train_runtime": 949.3813, | |
| "train_samples_per_second": 142.198, | |
| "train_steps_per_second": 1.109 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1053, | |
| "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": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.0638717415807058e+19, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |