| { | |
| "best_metric": 0.97, | |
| "best_model_checkpoint": "swin-tiny-patch4-window7-224-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": 5.58591365814209, | |
| "learning_rate": 4.716981132075472e-06, | |
| "loss": 2.3278, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05685856432125089, | |
| "grad_norm": 4.536377429962158, | |
| "learning_rate": 9.433962264150944e-06, | |
| "loss": 2.2763, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08528784648187633, | |
| "grad_norm": 4.974976539611816, | |
| "learning_rate": 1.4150943396226415e-05, | |
| "loss": 2.1632, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.11371712864250177, | |
| "grad_norm": 5.416134834289551, | |
| "learning_rate": 1.8867924528301888e-05, | |
| "loss": 1.9768, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.14214641080312723, | |
| "grad_norm": 8.11315631866455, | |
| "learning_rate": 2.358490566037736e-05, | |
| "loss": 1.7307, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17057569296375266, | |
| "grad_norm": 11.47766399383545, | |
| "learning_rate": 2.830188679245283e-05, | |
| "loss": 1.4492, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.19900497512437812, | |
| "grad_norm": 15.754049301147461, | |
| "learning_rate": 3.30188679245283e-05, | |
| "loss": 1.2237, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.22743425728500355, | |
| "grad_norm": 14.765463829040527, | |
| "learning_rate": 3.7735849056603776e-05, | |
| "loss": 1.0719, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.255863539445629, | |
| "grad_norm": 9.880644798278809, | |
| "learning_rate": 4.245283018867925e-05, | |
| "loss": 0.9904, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.28429282160625446, | |
| "grad_norm": 16.637893676757812, | |
| "learning_rate": 4.716981132075472e-05, | |
| "loss": 0.902, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.31272210376687987, | |
| "grad_norm": 14.136957168579102, | |
| "learning_rate": 4.978880675818374e-05, | |
| "loss": 0.8026, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3411513859275053, | |
| "grad_norm": 11.47761344909668, | |
| "learning_rate": 4.9260823653643085e-05, | |
| "loss": 0.7073, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3695806680881308, | |
| "grad_norm": 19.195457458496094, | |
| "learning_rate": 4.8732840549102435e-05, | |
| "loss": 0.7635, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.39800995024875624, | |
| "grad_norm": 13.571340560913086, | |
| "learning_rate": 4.820485744456177e-05, | |
| "loss": 0.7359, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.42643923240938164, | |
| "grad_norm": 9.327037811279297, | |
| "learning_rate": 4.767687434002112e-05, | |
| "loss": 0.6394, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4548685145700071, | |
| "grad_norm": 11.49142837524414, | |
| "learning_rate": 4.7148891235480466e-05, | |
| "loss": 0.6305, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.48329779673063256, | |
| "grad_norm": 9.735462188720703, | |
| "learning_rate": 4.662090813093981e-05, | |
| "loss": 0.6641, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.511727078891258, | |
| "grad_norm": 14.966803550720215, | |
| "learning_rate": 4.609292502639916e-05, | |
| "loss": 0.6013, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5401563610518835, | |
| "grad_norm": 10.155216217041016, | |
| "learning_rate": 4.55649419218585e-05, | |
| "loss": 0.6402, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5685856432125089, | |
| "grad_norm": 9.162310600280762, | |
| "learning_rate": 4.503695881731785e-05, | |
| "loss": 0.593, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5970149253731343, | |
| "grad_norm": 10.704963684082031, | |
| "learning_rate": 4.45089757127772e-05, | |
| "loss": 0.5944, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.6254442075337597, | |
| "grad_norm": 9.836397171020508, | |
| "learning_rate": 4.398099260823654e-05, | |
| "loss": 0.5976, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6538734896943852, | |
| "grad_norm": 9.258991241455078, | |
| "learning_rate": 4.3453009503695884e-05, | |
| "loss": 0.5715, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6823027718550106, | |
| "grad_norm": 10.040568351745605, | |
| "learning_rate": 4.292502639915523e-05, | |
| "loss": 0.6008, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7107320540156361, | |
| "grad_norm": 11.320852279663086, | |
| "learning_rate": 4.239704329461457e-05, | |
| "loss": 0.5517, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.7391613361762616, | |
| "grad_norm": 14.075531005859375, | |
| "learning_rate": 4.186906019007392e-05, | |
| "loss": 0.5207, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.767590618336887, | |
| "grad_norm": 8.05193042755127, | |
| "learning_rate": 4.1341077085533265e-05, | |
| "loss": 0.5601, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.7960199004975125, | |
| "grad_norm": 9.463001251220703, | |
| "learning_rate": 4.081309398099261e-05, | |
| "loss": 0.5246, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.8244491826581379, | |
| "grad_norm": 8.247632026672363, | |
| "learning_rate": 4.028511087645195e-05, | |
| "loss": 0.5047, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.8528784648187633, | |
| "grad_norm": 13.2966890335083, | |
| "learning_rate": 3.97571277719113e-05, | |
| "loss": 0.4996, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8813077469793887, | |
| "grad_norm": 9.029179573059082, | |
| "learning_rate": 3.9229144667370646e-05, | |
| "loss": 0.4836, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.9097370291400142, | |
| "grad_norm": 8.053582191467285, | |
| "learning_rate": 3.870116156282999e-05, | |
| "loss": 0.4933, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.9381663113006397, | |
| "grad_norm": 9.279789924621582, | |
| "learning_rate": 3.817317845828934e-05, | |
| "loss": 0.4748, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.9665955934612651, | |
| "grad_norm": 16.859342575073242, | |
| "learning_rate": 3.764519535374868e-05, | |
| "loss": 0.5047, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.9950248756218906, | |
| "grad_norm": 8.529033660888672, | |
| "learning_rate": 3.711721224920803e-05, | |
| "loss": 0.4857, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.997867803837953, | |
| "eval_accuracy": 0.947, | |
| "eval_loss": 0.158402681350708, | |
| "eval_runtime": 30.7764, | |
| "eval_samples_per_second": 162.462, | |
| "eval_steps_per_second": 5.101, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.023454157782516, | |
| "grad_norm": 11.528229713439941, | |
| "learning_rate": 3.658922914466738e-05, | |
| "loss": 0.4896, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.0518834399431414, | |
| "grad_norm": 10.65190315246582, | |
| "learning_rate": 3.6061246040126714e-05, | |
| "loss": 0.4846, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.080312722103767, | |
| "grad_norm": 9.108841896057129, | |
| "learning_rate": 3.5533262935586064e-05, | |
| "loss": 0.4785, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.1087420042643923, | |
| "grad_norm": 13.28585147857666, | |
| "learning_rate": 3.500527983104541e-05, | |
| "loss": 0.478, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.1371712864250179, | |
| "grad_norm": 12.707968711853027, | |
| "learning_rate": 3.447729672650475e-05, | |
| "loss": 0.4623, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.1656005685856432, | |
| "grad_norm": 10.01926040649414, | |
| "learning_rate": 3.3949313621964095e-05, | |
| "loss": 0.4082, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.1940298507462686, | |
| "grad_norm": 13.867918968200684, | |
| "learning_rate": 3.3421330517423445e-05, | |
| "loss": 0.4163, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.2224591329068941, | |
| "grad_norm": 8.311562538146973, | |
| "learning_rate": 3.289334741288279e-05, | |
| "loss": 0.4527, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.2508884150675195, | |
| "grad_norm": 8.391691207885742, | |
| "learning_rate": 3.236536430834213e-05, | |
| "loss": 0.4782, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.279317697228145, | |
| "grad_norm": 17.120820999145508, | |
| "learning_rate": 3.183738120380148e-05, | |
| "loss": 0.4233, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.3077469793887704, | |
| "grad_norm": 10.727948188781738, | |
| "learning_rate": 3.130939809926082e-05, | |
| "loss": 0.48, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.336176261549396, | |
| "grad_norm": 10.990976333618164, | |
| "learning_rate": 3.078141499472017e-05, | |
| "loss": 0.4643, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.3646055437100213, | |
| "grad_norm": 10.494908332824707, | |
| "learning_rate": 3.0253431890179517e-05, | |
| "loss": 0.4159, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.3930348258706466, | |
| "grad_norm": 12.02963638305664, | |
| "learning_rate": 2.972544878563886e-05, | |
| "loss": 0.4374, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.4214641080312722, | |
| "grad_norm": 9.365103721618652, | |
| "learning_rate": 2.9197465681098207e-05, | |
| "loss": 0.4491, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.4498933901918978, | |
| "grad_norm": 9.779677391052246, | |
| "learning_rate": 2.8669482576557548e-05, | |
| "loss": 0.4478, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.4783226723525231, | |
| "grad_norm": 10.054044723510742, | |
| "learning_rate": 2.8141499472016898e-05, | |
| "loss": 0.4301, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.5067519545131485, | |
| "grad_norm": 11.813431739807129, | |
| "learning_rate": 2.7613516367476245e-05, | |
| "loss": 0.4675, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.535181236673774, | |
| "grad_norm": 11.177428245544434, | |
| "learning_rate": 2.7085533262935585e-05, | |
| "loss": 0.4683, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.5636105188343994, | |
| "grad_norm": 8.778236389160156, | |
| "learning_rate": 2.6557550158394935e-05, | |
| "loss": 0.434, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.5920398009950247, | |
| "grad_norm": 9.267476081848145, | |
| "learning_rate": 2.6029567053854276e-05, | |
| "loss": 0.4076, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.6204690831556503, | |
| "grad_norm": 8.8627347946167, | |
| "learning_rate": 2.5501583949313622e-05, | |
| "loss": 0.441, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.6488983653162759, | |
| "grad_norm": 8.466072082519531, | |
| "learning_rate": 2.497360084477297e-05, | |
| "loss": 0.4463, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.6773276474769012, | |
| "grad_norm": 9.252815246582031, | |
| "learning_rate": 2.4445617740232313e-05, | |
| "loss": 0.4143, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.7057569296375266, | |
| "grad_norm": 11.29322624206543, | |
| "learning_rate": 2.391763463569166e-05, | |
| "loss": 0.4185, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.7341862117981521, | |
| "grad_norm": 8.104042053222656, | |
| "learning_rate": 2.3389651531151003e-05, | |
| "loss": 0.4053, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.7626154939587777, | |
| "grad_norm": 8.014995574951172, | |
| "learning_rate": 2.286166842661035e-05, | |
| "loss": 0.4331, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.7910447761194028, | |
| "grad_norm": 5.712684631347656, | |
| "learning_rate": 2.2333685322069694e-05, | |
| "loss": 0.366, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.8194740582800284, | |
| "grad_norm": 10.232897758483887, | |
| "learning_rate": 2.180570221752904e-05, | |
| "loss": 0.4036, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.847903340440654, | |
| "grad_norm": 7.411330699920654, | |
| "learning_rate": 2.1277719112988384e-05, | |
| "loss": 0.3692, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.8763326226012793, | |
| "grad_norm": 11.07670783996582, | |
| "learning_rate": 2.074973600844773e-05, | |
| "loss": 0.4312, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.9047619047619047, | |
| "grad_norm": 13.74502182006836, | |
| "learning_rate": 2.0221752903907075e-05, | |
| "loss": 0.3928, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.9331911869225302, | |
| "grad_norm": 8.972488403320312, | |
| "learning_rate": 1.9693769799366422e-05, | |
| "loss": 0.3961, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.9616204690831558, | |
| "grad_norm": 7.764671802520752, | |
| "learning_rate": 1.9165786694825765e-05, | |
| "loss": 0.3436, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.9900497512437811, | |
| "grad_norm": 14.391060829162598, | |
| "learning_rate": 1.863780359028511e-05, | |
| "loss": 0.3775, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.9985785358919688, | |
| "eval_accuracy": 0.9658, | |
| "eval_loss": 0.10344034433364868, | |
| "eval_runtime": 30.5926, | |
| "eval_samples_per_second": 163.438, | |
| "eval_steps_per_second": 5.132, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 2.0184790334044065, | |
| "grad_norm": 10.261659622192383, | |
| "learning_rate": 1.810982048574446e-05, | |
| "loss": 0.38, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.046908315565032, | |
| "grad_norm": 8.334637641906738, | |
| "learning_rate": 1.7581837381203803e-05, | |
| "loss": 0.3874, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.0753375977256576, | |
| "grad_norm": 10.484701156616211, | |
| "learning_rate": 1.7053854276663146e-05, | |
| "loss": 0.382, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.1037668798862827, | |
| "grad_norm": 7.803796291351318, | |
| "learning_rate": 1.6525871172122493e-05, | |
| "loss": 0.3841, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.1321961620469083, | |
| "grad_norm": 10.345030784606934, | |
| "learning_rate": 1.5997888067581837e-05, | |
| "loss": 0.3578, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.160625444207534, | |
| "grad_norm": 9.595276832580566, | |
| "learning_rate": 1.5469904963041184e-05, | |
| "loss": 0.3473, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.189054726368159, | |
| "grad_norm": 11.557988166809082, | |
| "learning_rate": 1.4941921858500529e-05, | |
| "loss": 0.3799, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.2174840085287846, | |
| "grad_norm": 7.893617153167725, | |
| "learning_rate": 1.4413938753959874e-05, | |
| "loss": 0.3934, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.24591329068941, | |
| "grad_norm": 10.705870628356934, | |
| "learning_rate": 1.388595564941922e-05, | |
| "loss": 0.38, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.2743425728500357, | |
| "grad_norm": 9.459859848022461, | |
| "learning_rate": 1.3357972544878563e-05, | |
| "loss": 0.3926, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.302771855010661, | |
| "grad_norm": 11.68864917755127, | |
| "learning_rate": 1.2829989440337912e-05, | |
| "loss": 0.3547, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.3312011371712864, | |
| "grad_norm": 10.39009952545166, | |
| "learning_rate": 1.2302006335797255e-05, | |
| "loss": 0.3512, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.359630419331912, | |
| "grad_norm": 8.366880416870117, | |
| "learning_rate": 1.17740232312566e-05, | |
| "loss": 0.4027, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.388059701492537, | |
| "grad_norm": 9.72803020477295, | |
| "learning_rate": 1.1246040126715946e-05, | |
| "loss": 0.3649, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.4164889836531627, | |
| "grad_norm": 11.347773551940918, | |
| "learning_rate": 1.0718057022175291e-05, | |
| "loss": 0.3598, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.4449182658137882, | |
| "grad_norm": 8.40634536743164, | |
| "learning_rate": 1.0190073917634636e-05, | |
| "loss": 0.3775, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.473347547974414, | |
| "grad_norm": 8.892584800720215, | |
| "learning_rate": 9.662090813093982e-06, | |
| "loss": 0.3607, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.501776830135039, | |
| "grad_norm": 14.059830665588379, | |
| "learning_rate": 9.134107708553327e-06, | |
| "loss": 0.3789, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.5302061122956645, | |
| "grad_norm": 9.594950675964355, | |
| "learning_rate": 8.606124604012672e-06, | |
| "loss": 0.3328, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.55863539445629, | |
| "grad_norm": 12.827249526977539, | |
| "learning_rate": 8.078141499472017e-06, | |
| "loss": 0.3698, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.587064676616915, | |
| "grad_norm": 12.365326881408691, | |
| "learning_rate": 7.5501583949313625e-06, | |
| "loss": 0.3644, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.6154939587775408, | |
| "grad_norm": 5.1415534019470215, | |
| "learning_rate": 7.022175290390708e-06, | |
| "loss": 0.3417, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.6439232409381663, | |
| "grad_norm": 7.832413673400879, | |
| "learning_rate": 6.494192185850054e-06, | |
| "loss": 0.3696, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.672352523098792, | |
| "grad_norm": 8.78602123260498, | |
| "learning_rate": 5.966209081309398e-06, | |
| "loss": 0.3428, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.7007818052594175, | |
| "grad_norm": 9.317280769348145, | |
| "learning_rate": 5.438225976768744e-06, | |
| "loss": 0.3492, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.7292110874200426, | |
| "grad_norm": 7.861667633056641, | |
| "learning_rate": 4.910242872228089e-06, | |
| "loss": 0.3556, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.757640369580668, | |
| "grad_norm": 8.467522621154785, | |
| "learning_rate": 4.382259767687434e-06, | |
| "loss": 0.3456, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.7860696517412933, | |
| "grad_norm": 10.145249366760254, | |
| "learning_rate": 3.854276663146779e-06, | |
| "loss": 0.3221, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.814498933901919, | |
| "grad_norm": 11.336479187011719, | |
| "learning_rate": 3.326293558606125e-06, | |
| "loss": 0.4209, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.8429282160625444, | |
| "grad_norm": 9.06911849975586, | |
| "learning_rate": 2.79831045406547e-06, | |
| "loss": 0.3384, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.87135749822317, | |
| "grad_norm": 8.496155738830566, | |
| "learning_rate": 2.2703273495248154e-06, | |
| "loss": 0.3254, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.8997867803837956, | |
| "grad_norm": 7.5788960456848145, | |
| "learning_rate": 1.7423442449841606e-06, | |
| "loss": 0.3787, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.9282160625444207, | |
| "grad_norm": 11.145151138305664, | |
| "learning_rate": 1.2143611404435059e-06, | |
| "loss": 0.3311, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.9566453447050463, | |
| "grad_norm": 6.017477512359619, | |
| "learning_rate": 6.863780359028511e-07, | |
| "loss": 0.3586, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.9850746268656714, | |
| "grad_norm": 12.3096342086792, | |
| "learning_rate": 1.5839493136219642e-07, | |
| "loss": 0.3216, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.9936034115138592, | |
| "eval_accuracy": 0.97, | |
| "eval_loss": 0.0925411656498909, | |
| "eval_runtime": 30.7915, | |
| "eval_samples_per_second": 162.382, | |
| "eval_steps_per_second": 5.099, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 2.9936034115138592, | |
| "step": 1053, | |
| "total_flos": 3.3497451642252165e+18, | |
| "train_loss": 0.5581566964798843, | |
| "train_runtime": 2011.3946, | |
| "train_samples_per_second": 67.118, | |
| "train_steps_per_second": 0.524 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1053, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 3.3497451642252165e+18, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |