|
{ |
|
"best_metric": 0.9844444444444445, |
|
"best_model_checkpoint": "vit-base-patch16-224-in21k-finetuned-eurosat/checkpoint-2277", |
|
"epoch": 3.0, |
|
"global_step": 2277, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 2.1929824561403507e-06, |
|
"loss": 2.3182, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.3859649122807014e-06, |
|
"loss": 2.3088, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 6.578947368421053e-06, |
|
"loss": 2.2658, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 8.771929824561403e-06, |
|
"loss": 2.1543, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 1.0964912280701754e-05, |
|
"loss": 2.2097, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1.3157894736842106e-05, |
|
"loss": 2.1236, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 1.5350877192982457e-05, |
|
"loss": 1.9738, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 1.7543859649122806e-05, |
|
"loss": 1.8933, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9736842105263158e-05, |
|
"loss": 1.7186, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 2.1929824561403507e-05, |
|
"loss": 1.7101, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 2.412280701754386e-05, |
|
"loss": 1.7188, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 2.6315789473684212e-05, |
|
"loss": 1.0406, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 2.850877192982456e-05, |
|
"loss": 1.1604, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 3.0701754385964913e-05, |
|
"loss": 1.1226, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 3.289473684210527e-05, |
|
"loss": 1.0185, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 3.508771929824561e-05, |
|
"loss": 1.0203, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 3.728070175438597e-05, |
|
"loss": 0.5058, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 3.9473684210526316e-05, |
|
"loss": 0.7132, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 4.166666666666667e-05, |
|
"loss": 0.7988, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 4.3859649122807014e-05, |
|
"loss": 0.5992, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.605263157894737e-05, |
|
"loss": 0.7898, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 4.824561403508772e-05, |
|
"loss": 0.6248, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 4.9951195705222064e-05, |
|
"loss": 0.6488, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 4.970717423133236e-05, |
|
"loss": 0.6448, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 4.946315275744266e-05, |
|
"loss": 0.6157, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 4.9219131283552955e-05, |
|
"loss": 0.636, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.897510980966325e-05, |
|
"loss": 0.6592, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 4.873108833577355e-05, |
|
"loss": 0.6987, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 4.848706686188385e-05, |
|
"loss": 0.5183, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 4.824304538799414e-05, |
|
"loss": 0.2226, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 4.799902391410445e-05, |
|
"loss": 1.0035, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 4.7755002440214744e-05, |
|
"loss": 0.2608, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 4.751098096632504e-05, |
|
"loss": 0.2487, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 4.726695949243534e-05, |
|
"loss": 0.5035, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.7022938018545635e-05, |
|
"loss": 0.6245, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.677891654465593e-05, |
|
"loss": 0.2621, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.6534895070766235e-05, |
|
"loss": 0.2333, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.6290873596876525e-05, |
|
"loss": 0.1616, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.604685212298683e-05, |
|
"loss": 0.2174, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 4.580283064909712e-05, |
|
"loss": 0.5526, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.555880917520742e-05, |
|
"loss": 0.2229, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 4.531478770131772e-05, |
|
"loss": 0.1677, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 4.507076622742802e-05, |
|
"loss": 0.14, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 4.4826744753538314e-05, |
|
"loss": 0.1421, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 4.458272327964861e-05, |
|
"loss": 0.1422, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 4.433870180575891e-05, |
|
"loss": 0.1693, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 4.409468033186921e-05, |
|
"loss": 0.1312, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 4.38506588579795e-05, |
|
"loss": 0.5159, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 4.3606637384089805e-05, |
|
"loss": 0.1083, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 4.33626159102001e-05, |
|
"loss": 0.5269, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 4.31185944363104e-05, |
|
"loss": 0.1268, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 4.2874572962420696e-05, |
|
"loss": 0.0972, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.263055148853099e-05, |
|
"loss": 0.104, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.238653001464129e-05, |
|
"loss": 0.3534, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.2142508540751594e-05, |
|
"loss": 0.1813, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 4.1898487066861884e-05, |
|
"loss": 0.2723, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 4.165446559297219e-05, |
|
"loss": 0.1537, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 4.141044411908248e-05, |
|
"loss": 0.5187, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 4.116642264519278e-05, |
|
"loss": 0.1365, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 4.092240117130308e-05, |
|
"loss": 0.2843, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 4.0678379697413375e-05, |
|
"loss": 0.385, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 4.043435822352367e-05, |
|
"loss": 0.3066, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 4.019033674963397e-05, |
|
"loss": 0.1013, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 3.9946315275744266e-05, |
|
"loss": 0.5619, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 3.970229380185457e-05, |
|
"loss": 0.1353, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 3.945827232796486e-05, |
|
"loss": 0.2441, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 3.9214250854075164e-05, |
|
"loss": 0.091, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 3.897022938018546e-05, |
|
"loss": 0.1073, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 3.872620790629576e-05, |
|
"loss": 0.1118, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.8482186432406055e-05, |
|
"loss": 0.0803, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 3.823816495851635e-05, |
|
"loss": 0.127, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 3.799414348462665e-05, |
|
"loss": 0.109, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 3.775012201073695e-05, |
|
"loss": 0.0735, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 3.750610053684724e-05, |
|
"loss": 0.1187, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 3.7262079062957546e-05, |
|
"loss": 0.2969, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_accuracy": 0.9785185185185186, |
|
"eval_loss": 0.1263427734375, |
|
"eval_runtime": 10.5602, |
|
"eval_samples_per_second": 255.677, |
|
"eval_steps_per_second": 63.919, |
|
"step": 759 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 3.7018057589067836e-05, |
|
"loss": 0.2753, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 3.677403611517814e-05, |
|
"loss": 0.403, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 3.653001464128844e-05, |
|
"loss": 0.2884, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 3.6285993167398734e-05, |
|
"loss": 0.4866, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 3.604197169350903e-05, |
|
"loss": 0.4529, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 3.579795021961933e-05, |
|
"loss": 0.063, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 3.5553928745729625e-05, |
|
"loss": 0.0952, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 3.530990727183993e-05, |
|
"loss": 0.0886, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 3.506588579795022e-05, |
|
"loss": 0.2751, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 3.482186432406052e-05, |
|
"loss": 0.09, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 3.457784285017082e-05, |
|
"loss": 0.8028, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 3.4333821376281116e-05, |
|
"loss": 0.0623, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 3.408979990239141e-05, |
|
"loss": 0.0583, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 3.384577842850171e-05, |
|
"loss": 0.5004, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 3.360175695461201e-05, |
|
"loss": 0.0587, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 3.335773548072231e-05, |
|
"loss": 0.066, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 3.31137140068326e-05, |
|
"loss": 0.1028, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 3.2869692532942905e-05, |
|
"loss": 0.069, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 3.2625671059053195e-05, |
|
"loss": 0.1084, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 3.23816495851635e-05, |
|
"loss": 0.8451, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 3.2137628111273795e-05, |
|
"loss": 0.0993, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 3.189360663738409e-05, |
|
"loss": 0.0488, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 3.164958516349439e-05, |
|
"loss": 0.0645, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 3.1405563689604686e-05, |
|
"loss": 0.0586, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 3.116154221571498e-05, |
|
"loss": 0.4479, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 3.091752074182529e-05, |
|
"loss": 0.4036, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 3.067349926793558e-05, |
|
"loss": 0.0975, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 3.042947779404588e-05, |
|
"loss": 0.7869, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 3.0185456320156174e-05, |
|
"loss": 0.0552, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 2.9941434846266475e-05, |
|
"loss": 0.0515, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 2.969741337237677e-05, |
|
"loss": 0.6272, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 2.9453391898487072e-05, |
|
"loss": 0.0522, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 2.9209370424597365e-05, |
|
"loss": 0.0493, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 2.8965348950707666e-05, |
|
"loss": 0.4377, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 2.872132747681796e-05, |
|
"loss": 0.0417, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 2.847730600292826e-05, |
|
"loss": 0.0684, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 2.8233284529038557e-05, |
|
"loss": 0.0735, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 1.49, |
|
"learning_rate": 2.7989263055148857e-05, |
|
"loss": 0.0614, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 2.774524158125915e-05, |
|
"loss": 0.3421, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 2.750122010736945e-05, |
|
"loss": 0.104, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 2.7257198633479748e-05, |
|
"loss": 0.6174, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 2.7013177159590048e-05, |
|
"loss": 0.1763, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 2.676915568570034e-05, |
|
"loss": 0.2104, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 2.6525134211810642e-05, |
|
"loss": 0.0592, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 2.628111273792094e-05, |
|
"loss": 0.1282, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 2.603709126403124e-05, |
|
"loss": 0.0435, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 2.5793069790141533e-05, |
|
"loss": 0.2258, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 2.5549048316251833e-05, |
|
"loss": 0.2802, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 2.530502684236213e-05, |
|
"loss": 0.0463, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 2.506100536847243e-05, |
|
"loss": 0.1848, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 2.4816983894582724e-05, |
|
"loss": 0.0794, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 2.457296242069302e-05, |
|
"loss": 0.0435, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 2.4328940946803318e-05, |
|
"loss": 0.0921, |
|
"step": 1280 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 2.4084919472913618e-05, |
|
"loss": 0.1854, |
|
"step": 1290 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 2.3840897999023915e-05, |
|
"loss": 0.0384, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 2.3596876525134212e-05, |
|
"loss": 0.0462, |
|
"step": 1310 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 2.335285505124451e-05, |
|
"loss": 0.0802, |
|
"step": 1320 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 2.310883357735481e-05, |
|
"loss": 0.046, |
|
"step": 1330 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 2.2864812103465106e-05, |
|
"loss": 0.0775, |
|
"step": 1340 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 2.2620790629575403e-05, |
|
"loss": 0.0532, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 2.23767691556857e-05, |
|
"loss": 0.0615, |
|
"step": 1360 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 2.2132747681795997e-05, |
|
"loss": 0.0435, |
|
"step": 1370 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 2.1888726207906297e-05, |
|
"loss": 0.0372, |
|
"step": 1380 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 2.1644704734016594e-05, |
|
"loss": 0.0351, |
|
"step": 1390 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 2.140068326012689e-05, |
|
"loss": 0.0523, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 2.1156661786237188e-05, |
|
"loss": 0.0345, |
|
"step": 1410 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 2.0912640312347485e-05, |
|
"loss": 0.0359, |
|
"step": 1420 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 2.0668618838457785e-05, |
|
"loss": 0.5072, |
|
"step": 1430 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 2.0424597364568082e-05, |
|
"loss": 0.0281, |
|
"step": 1440 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"learning_rate": 2.018057589067838e-05, |
|
"loss": 0.0366, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 1.9936554416788676e-05, |
|
"loss": 0.2221, |
|
"step": 1460 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 1.9692532942898977e-05, |
|
"loss": 0.0413, |
|
"step": 1470 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 1.9448511469009274e-05, |
|
"loss": 0.3138, |
|
"step": 1480 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 1.920448999511957e-05, |
|
"loss": 0.113, |
|
"step": 1490 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 1.8960468521229867e-05, |
|
"loss": 0.037, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 1.8716447047340164e-05, |
|
"loss": 0.0375, |
|
"step": 1510 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.9792592592592593, |
|
"eval_loss": 0.09527587890625, |
|
"eval_runtime": 9.7577, |
|
"eval_samples_per_second": 276.704, |
|
"eval_steps_per_second": 69.176, |
|
"step": 1518 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 1.8472425573450465e-05, |
|
"loss": 0.0722, |
|
"step": 1520 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 1.822840409956076e-05, |
|
"loss": 0.0311, |
|
"step": 1530 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 1.798438262567106e-05, |
|
"loss": 0.4362, |
|
"step": 1540 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 1.7740361151781356e-05, |
|
"loss": 0.1499, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 1.7496339677891656e-05, |
|
"loss": 0.2203, |
|
"step": 1560 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 1.7252318204001953e-05, |
|
"loss": 0.0499, |
|
"step": 1570 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 1.700829673011225e-05, |
|
"loss": 0.1047, |
|
"step": 1580 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 1.6764275256222547e-05, |
|
"loss": 0.0313, |
|
"step": 1590 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 1.6520253782332844e-05, |
|
"loss": 0.2069, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 2.12, |
|
"learning_rate": 1.6276232308443144e-05, |
|
"loss": 0.0337, |
|
"step": 1610 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 1.603221083455344e-05, |
|
"loss": 0.1789, |
|
"step": 1620 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 1.5788189360663738e-05, |
|
"loss": 0.0437, |
|
"step": 1630 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 1.5544167886774035e-05, |
|
"loss": 0.0346, |
|
"step": 1640 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 1.5300146412884335e-05, |
|
"loss": 0.0351, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.5056124938994632e-05, |
|
"loss": 0.2496, |
|
"step": 1660 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.4812103465104929e-05, |
|
"loss": 0.1546, |
|
"step": 1670 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.4568081991215226e-05, |
|
"loss": 0.6364, |
|
"step": 1680 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.4324060517325525e-05, |
|
"loss": 0.0405, |
|
"step": 1690 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.4080039043435821e-05, |
|
"loss": 0.083, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.383601756954612e-05, |
|
"loss": 0.0354, |
|
"step": 1710 |
|
}, |
|
{ |
|
"epoch": 2.27, |
|
"learning_rate": 1.3591996095656417e-05, |
|
"loss": 0.0658, |
|
"step": 1720 |
|
}, |
|
{ |
|
"epoch": 2.28, |
|
"learning_rate": 1.3347974621766716e-05, |
|
"loss": 0.09, |
|
"step": 1730 |
|
}, |
|
{ |
|
"epoch": 2.29, |
|
"learning_rate": 1.3103953147877013e-05, |
|
"loss": 0.0259, |
|
"step": 1740 |
|
}, |
|
{ |
|
"epoch": 2.31, |
|
"learning_rate": 1.2859931673987311e-05, |
|
"loss": 0.041, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 2.32, |
|
"learning_rate": 1.2615910200097608e-05, |
|
"loss": 0.1706, |
|
"step": 1760 |
|
}, |
|
{ |
|
"epoch": 2.33, |
|
"learning_rate": 1.2371888726207907e-05, |
|
"loss": 0.0301, |
|
"step": 1770 |
|
}, |
|
{ |
|
"epoch": 2.35, |
|
"learning_rate": 1.2127867252318205e-05, |
|
"loss": 0.2378, |
|
"step": 1780 |
|
}, |
|
{ |
|
"epoch": 2.36, |
|
"learning_rate": 1.1883845778428502e-05, |
|
"loss": 0.0277, |
|
"step": 1790 |
|
}, |
|
{ |
|
"epoch": 2.37, |
|
"learning_rate": 1.1639824304538801e-05, |
|
"loss": 0.0265, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 2.38, |
|
"learning_rate": 1.1395802830649098e-05, |
|
"loss": 0.036, |
|
"step": 1810 |
|
}, |
|
{ |
|
"epoch": 2.4, |
|
"learning_rate": 1.1151781356759395e-05, |
|
"loss": 0.0587, |
|
"step": 1820 |
|
}, |
|
{ |
|
"epoch": 2.41, |
|
"learning_rate": 1.0907759882869694e-05, |
|
"loss": 0.0388, |
|
"step": 1830 |
|
}, |
|
{ |
|
"epoch": 2.42, |
|
"learning_rate": 1.066373840897999e-05, |
|
"loss": 0.0492, |
|
"step": 1840 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 1.0419716935090289e-05, |
|
"loss": 0.0276, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 2.45, |
|
"learning_rate": 1.0175695461200586e-05, |
|
"loss": 0.0286, |
|
"step": 1860 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 9.931673987310885e-06, |
|
"loss": 0.0285, |
|
"step": 1870 |
|
}, |
|
{ |
|
"epoch": 2.48, |
|
"learning_rate": 9.687652513421182e-06, |
|
"loss": 0.0447, |
|
"step": 1880 |
|
}, |
|
{ |
|
"epoch": 2.49, |
|
"learning_rate": 9.44363103953148e-06, |
|
"loss": 0.0258, |
|
"step": 1890 |
|
}, |
|
{ |
|
"epoch": 2.5, |
|
"learning_rate": 9.199609565641777e-06, |
|
"loss": 0.0248, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 2.52, |
|
"learning_rate": 8.955588091752074e-06, |
|
"loss": 0.0379, |
|
"step": 1910 |
|
}, |
|
{ |
|
"epoch": 2.53, |
|
"learning_rate": 8.711566617862373e-06, |
|
"loss": 0.0386, |
|
"step": 1920 |
|
}, |
|
{ |
|
"epoch": 2.54, |
|
"learning_rate": 8.46754514397267e-06, |
|
"loss": 0.0938, |
|
"step": 1930 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 8.223523670082968e-06, |
|
"loss": 0.0381, |
|
"step": 1940 |
|
}, |
|
{ |
|
"epoch": 2.57, |
|
"learning_rate": 7.979502196193265e-06, |
|
"loss": 0.0589, |
|
"step": 1950 |
|
}, |
|
{ |
|
"epoch": 2.58, |
|
"learning_rate": 7.735480722303564e-06, |
|
"loss": 0.3975, |
|
"step": 1960 |
|
}, |
|
{ |
|
"epoch": 2.6, |
|
"learning_rate": 7.491459248413861e-06, |
|
"loss": 0.0307, |
|
"step": 1970 |
|
}, |
|
{ |
|
"epoch": 2.61, |
|
"learning_rate": 7.247437774524159e-06, |
|
"loss": 0.046, |
|
"step": 1980 |
|
}, |
|
{ |
|
"epoch": 2.62, |
|
"learning_rate": 7.0034163006344565e-06, |
|
"loss": 0.026, |
|
"step": 1990 |
|
}, |
|
{ |
|
"epoch": 2.64, |
|
"learning_rate": 6.759394826744754e-06, |
|
"loss": 0.0749, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 2.65, |
|
"learning_rate": 6.515373352855052e-06, |
|
"loss": 0.0306, |
|
"step": 2010 |
|
}, |
|
{ |
|
"epoch": 2.66, |
|
"learning_rate": 6.27135187896535e-06, |
|
"loss": 0.0536, |
|
"step": 2020 |
|
}, |
|
{ |
|
"epoch": 2.67, |
|
"learning_rate": 6.027330405075647e-06, |
|
"loss": 0.0265, |
|
"step": 2030 |
|
}, |
|
{ |
|
"epoch": 2.69, |
|
"learning_rate": 5.7833089311859446e-06, |
|
"loss": 0.0329, |
|
"step": 2040 |
|
}, |
|
{ |
|
"epoch": 2.7, |
|
"learning_rate": 5.539287457296242e-06, |
|
"loss": 0.5737, |
|
"step": 2050 |
|
}, |
|
{ |
|
"epoch": 2.71, |
|
"learning_rate": 5.29526598340654e-06, |
|
"loss": 0.8366, |
|
"step": 2060 |
|
}, |
|
{ |
|
"epoch": 2.73, |
|
"learning_rate": 5.051244509516838e-06, |
|
"loss": 0.0282, |
|
"step": 2070 |
|
}, |
|
{ |
|
"epoch": 2.74, |
|
"learning_rate": 4.807223035627136e-06, |
|
"loss": 0.0262, |
|
"step": 2080 |
|
}, |
|
{ |
|
"epoch": 2.75, |
|
"learning_rate": 4.5632015617374335e-06, |
|
"loss": 0.1776, |
|
"step": 2090 |
|
}, |
|
{ |
|
"epoch": 2.77, |
|
"learning_rate": 4.319180087847731e-06, |
|
"loss": 0.024, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 2.78, |
|
"learning_rate": 4.075158613958028e-06, |
|
"loss": 0.1574, |
|
"step": 2110 |
|
}, |
|
{ |
|
"epoch": 2.79, |
|
"learning_rate": 3.831137140068326e-06, |
|
"loss": 0.1553, |
|
"step": 2120 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"learning_rate": 3.587115666178624e-06, |
|
"loss": 0.1287, |
|
"step": 2130 |
|
}, |
|
{ |
|
"epoch": 2.82, |
|
"learning_rate": 3.3430941922889216e-06, |
|
"loss": 0.2801, |
|
"step": 2140 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"learning_rate": 3.0990727183992194e-06, |
|
"loss": 0.0326, |
|
"step": 2150 |
|
}, |
|
{ |
|
"epoch": 2.85, |
|
"learning_rate": 2.855051244509517e-06, |
|
"loss": 0.2817, |
|
"step": 2160 |
|
}, |
|
{ |
|
"epoch": 2.86, |
|
"learning_rate": 2.611029770619815e-06, |
|
"loss": 0.0579, |
|
"step": 2170 |
|
}, |
|
{ |
|
"epoch": 2.87, |
|
"learning_rate": 2.3670082967301123e-06, |
|
"loss": 0.0243, |
|
"step": 2180 |
|
}, |
|
{ |
|
"epoch": 2.89, |
|
"learning_rate": 2.12298682284041e-06, |
|
"loss": 0.0426, |
|
"step": 2190 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"learning_rate": 1.8789653489507079e-06, |
|
"loss": 0.1296, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 2.91, |
|
"learning_rate": 1.6349438750610055e-06, |
|
"loss": 0.5111, |
|
"step": 2210 |
|
}, |
|
{ |
|
"epoch": 2.92, |
|
"learning_rate": 1.3909224011713032e-06, |
|
"loss": 0.1463, |
|
"step": 2220 |
|
}, |
|
{ |
|
"epoch": 2.94, |
|
"learning_rate": 1.1469009272816008e-06, |
|
"loss": 0.05, |
|
"step": 2230 |
|
}, |
|
{ |
|
"epoch": 2.95, |
|
"learning_rate": 9.028794533918985e-07, |
|
"loss": 0.0225, |
|
"step": 2240 |
|
}, |
|
{ |
|
"epoch": 2.96, |
|
"learning_rate": 6.588579795021963e-07, |
|
"loss": 0.0311, |
|
"step": 2250 |
|
}, |
|
{ |
|
"epoch": 2.98, |
|
"learning_rate": 4.148365056124939e-07, |
|
"loss": 0.2809, |
|
"step": 2260 |
|
}, |
|
{ |
|
"epoch": 2.99, |
|
"learning_rate": 1.7081503172279163e-07, |
|
"loss": 0.0345, |
|
"step": 2270 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_accuracy": 0.9844444444444445, |
|
"eval_loss": 0.07232666015625, |
|
"eval_runtime": 9.9115, |
|
"eval_samples_per_second": 272.41, |
|
"eval_steps_per_second": 68.102, |
|
"step": 2277 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 2277, |
|
"total_flos": 1.8069699791226824e+20, |
|
"train_loss": 0.3087928044884991, |
|
"train_runtime": 984.02, |
|
"train_samples_per_second": 74.047, |
|
"train_steps_per_second": 2.314 |
|
} |
|
], |
|
"max_steps": 2277, |
|
"num_train_epochs": 3, |
|
"start_time": 1668807038.3244002, |
|
"total_flos": 1.8069699791226824e+20, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|