| { | |
| "best_metric": 0.010264705866575241, | |
| "best_model_checkpoint": "./outputs/checkpoint-3320", | |
| "epoch": 5.0, | |
| "global_step": 3320, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.993975903614458e-05, | |
| "loss": 0.4277, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.987951807228916e-05, | |
| "loss": 0.1348, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9819277108433738e-05, | |
| "loss": 0.0648, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9759036144578312e-05, | |
| "loss": 0.0249, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9698795180722894e-05, | |
| "loss": 0.0311, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.963855421686747e-05, | |
| "loss": 0.0213, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.957831325301205e-05, | |
| "loss": 0.0405, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.951807228915663e-05, | |
| "loss": 0.0216, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9457831325301208e-05, | |
| "loss": 0.1032, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9397590361445785e-05, | |
| "loss": 0.0347, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.9337349397590363e-05, | |
| "loss": 0.0211, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.927710843373494e-05, | |
| "loss": 0.082, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.921686746987952e-05, | |
| "loss": 0.025, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.91566265060241e-05, | |
| "loss": 0.0109, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.9096385542168677e-05, | |
| "loss": 0.0142, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9036144578313255e-05, | |
| "loss": 0.0696, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.8975903614457833e-05, | |
| "loss": 0.0536, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.891566265060241e-05, | |
| "loss": 0.0538, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.8855421686746988e-05, | |
| "loss": 0.0688, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.879518072289157e-05, | |
| "loss": 0.0319, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.8734939759036147e-05, | |
| "loss": 0.0592, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.8674698795180725e-05, | |
| "loss": 0.0252, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.8614457831325302e-05, | |
| "loss": 0.0031, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.855421686746988e-05, | |
| "loss": 0.0124, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.8493975903614458e-05, | |
| "loss": 0.0498, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.843373493975904e-05, | |
| "loss": 0.0222, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.8373493975903617e-05, | |
| "loss": 0.0392, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.8313253012048194e-05, | |
| "loss": 0.0218, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.8253012048192772e-05, | |
| "loss": 0.0247, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.819277108433735e-05, | |
| "loss": 0.0471, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.8132530120481927e-05, | |
| "loss": 0.0162, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.807228915662651e-05, | |
| "loss": 0.053, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.8012048192771086e-05, | |
| "loss": 0.0308, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.7951807228915664e-05, | |
| "loss": 0.0771, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.789156626506024e-05, | |
| "loss": 0.011, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.783132530120482e-05, | |
| "loss": 0.0141, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.7771084337349397e-05, | |
| "loss": 0.0679, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.7710843373493978e-05, | |
| "loss": 0.036, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.7650602409638556e-05, | |
| "loss": 0.0213, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.7590361445783134e-05, | |
| "loss": 0.0456, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.753012048192771e-05, | |
| "loss": 0.0506, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.746987951807229e-05, | |
| "loss": 0.0335, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.740963855421687e-05, | |
| "loss": 0.0386, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.7349397590361448e-05, | |
| "loss": 0.0787, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.7289156626506026e-05, | |
| "loss": 0.0282, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.7228915662650603e-05, | |
| "loss": 0.0165, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.716867469879518e-05, | |
| "loss": 0.1302, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.710843373493976e-05, | |
| "loss": 0.0564, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.704819277108434e-05, | |
| "loss": 0.0049, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.6987951807228917e-05, | |
| "loss": 0.0019, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.6927710843373495e-05, | |
| "loss": 0.0084, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.6867469879518076e-05, | |
| "loss": 0.0045, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.680722891566265e-05, | |
| "loss": 0.0362, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.6746987951807228e-05, | |
| "loss": 0.0075, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.668674698795181e-05, | |
| "loss": 0.0434, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6626506024096387e-05, | |
| "loss": 0.0154, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.6566265060240965e-05, | |
| "loss": 0.004, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.6506024096385546e-05, | |
| "loss": 0.0223, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.6445783132530124e-05, | |
| "loss": 0.1085, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.63855421686747e-05, | |
| "loss": 0.0304, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.632530120481928e-05, | |
| "loss": 0.0969, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.6265060240963857e-05, | |
| "loss": 0.0237, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.6204819277108434e-05, | |
| "loss": 0.0444, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.6144578313253015e-05, | |
| "loss": 0.0385, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.6084337349397593e-05, | |
| "loss": 0.058, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.602409638554217e-05, | |
| "loss": 0.0072, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9973333333333333, | |
| "eval_loss": 0.014009708538651466, | |
| "eval_runtime": 66.9467, | |
| "eval_samples_per_second": 56.015, | |
| "eval_steps_per_second": 3.51, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.596385542168675e-05, | |
| "loss": 0.0736, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.5903614457831326e-05, | |
| "loss": 0.0352, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.5843373493975904e-05, | |
| "loss": 0.0145, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.5783132530120485e-05, | |
| "loss": 0.0691, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.5722891566265063e-05, | |
| "loss": 0.0248, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.566265060240964e-05, | |
| "loss": 0.0399, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.5602409638554218e-05, | |
| "loss": 0.0286, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.5542168674698796e-05, | |
| "loss": 0.0299, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.5481927710843374e-05, | |
| "loss": 0.0592, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.5421686746987955e-05, | |
| "loss": 0.0385, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.5361445783132532e-05, | |
| "loss": 0.0441, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.530120481927711e-05, | |
| "loss": 0.0177, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.524096385542169e-05, | |
| "loss": 0.0297, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.5180722891566266e-05, | |
| "loss": 0.0244, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.5120481927710843e-05, | |
| "loss": 0.0275, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.5060240963855424e-05, | |
| "loss": 0.0359, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.0183, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.4939759036144578e-05, | |
| "loss": 0.0323, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.487951807228916e-05, | |
| "loss": 0.0182, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.4819277108433737e-05, | |
| "loss": 0.0474, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.4759036144578313e-05, | |
| "loss": 0.0516, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.4698795180722894e-05, | |
| "loss": 0.0614, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.4638554216867472e-05, | |
| "loss": 0.0214, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.457831325301205e-05, | |
| "loss": 0.081, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.4518072289156629e-05, | |
| "loss": 0.0384, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.4457831325301207e-05, | |
| "loss": 0.0333, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.4397590361445784e-05, | |
| "loss": 0.025, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.4337349397590364e-05, | |
| "loss": 0.0173, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4277108433734941e-05, | |
| "loss": 0.046, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.4216867469879519e-05, | |
| "loss": 0.0204, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.4156626506024098e-05, | |
| "loss": 0.0235, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.4096385542168676e-05, | |
| "loss": 0.0601, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.4036144578313254e-05, | |
| "loss": 0.0323, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.3975903614457833e-05, | |
| "loss": 0.0101, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.3915662650602411e-05, | |
| "loss": 0.051, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 1.3855421686746989e-05, | |
| "loss": 0.0088, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.3795180722891568e-05, | |
| "loss": 0.029, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1.3734939759036146e-05, | |
| "loss": 0.049, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.3674698795180723e-05, | |
| "loss": 0.0292, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.3614457831325303e-05, | |
| "loss": 0.0395, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 1.355421686746988e-05, | |
| "loss": 0.0981, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.3493975903614458e-05, | |
| "loss": 0.0288, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.3433734939759038e-05, | |
| "loss": 0.0216, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 1.3373493975903615e-05, | |
| "loss": 0.0037, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.3313253012048193e-05, | |
| "loss": 0.0344, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.3253012048192772e-05, | |
| "loss": 0.0576, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.319277108433735e-05, | |
| "loss": 0.0586, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.3132530120481928e-05, | |
| "loss": 0.0425, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.3072289156626507e-05, | |
| "loss": 0.0243, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.3012048192771085e-05, | |
| "loss": 0.0427, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.2951807228915663e-05, | |
| "loss": 0.0662, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.2891566265060242e-05, | |
| "loss": 0.0323, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.283132530120482e-05, | |
| "loss": 0.0523, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.2771084337349398e-05, | |
| "loss": 0.0209, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.2710843373493977e-05, | |
| "loss": 0.0192, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.2650602409638555e-05, | |
| "loss": 0.0094, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.2590361445783132e-05, | |
| "loss": 0.0307, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.2530120481927712e-05, | |
| "loss": 0.0383, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.246987951807229e-05, | |
| "loss": 0.0464, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.2409638554216869e-05, | |
| "loss": 0.0346, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.2349397590361447e-05, | |
| "loss": 0.0393, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.2289156626506024e-05, | |
| "loss": 0.0491, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.2228915662650604e-05, | |
| "loss": 0.0299, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.2168674698795181e-05, | |
| "loss": 0.0276, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.2108433734939759e-05, | |
| "loss": 0.0105, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.204819277108434e-05, | |
| "loss": 0.0062, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9965333333333334, | |
| "eval_loss": 0.011143543757498264, | |
| "eval_runtime": 67.068, | |
| "eval_samples_per_second": 55.913, | |
| "eval_steps_per_second": 3.504, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.1987951807228916e-05, | |
| "loss": 0.0274, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.1927710843373494e-05, | |
| "loss": 0.0515, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.1867469879518075e-05, | |
| "loss": 0.0031, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.1807228915662651e-05, | |
| "loss": 0.0072, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.1746987951807229e-05, | |
| "loss": 0.0791, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.168674698795181e-05, | |
| "loss": 0.0173, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.1626506024096388e-05, | |
| "loss": 0.0039, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.1566265060240964e-05, | |
| "loss": 0.0284, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.1506024096385545e-05, | |
| "loss": 0.0016, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.1445783132530122e-05, | |
| "loss": 0.0002, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.13855421686747e-05, | |
| "loss": 0.0336, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.132530120481928e-05, | |
| "loss": 0.0318, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.1265060240963857e-05, | |
| "loss": 0.0398, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.1204819277108435e-05, | |
| "loss": 0.0206, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.1144578313253014e-05, | |
| "loss": 0.0336, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.1084337349397592e-05, | |
| "loss": 0.0148, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.102409638554217e-05, | |
| "loss": 0.0062, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.0963855421686749e-05, | |
| "loss": 0.0496, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.0903614457831327e-05, | |
| "loss": 0.0147, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.0843373493975904e-05, | |
| "loss": 0.0173, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.0783132530120484e-05, | |
| "loss": 0.006, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.0722891566265062e-05, | |
| "loss": 0.0454, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.066265060240964e-05, | |
| "loss": 0.0198, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0602409638554219e-05, | |
| "loss": 0.0497, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0542168674698796e-05, | |
| "loss": 0.0127, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0481927710843374e-05, | |
| "loss": 0.0356, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0421686746987953e-05, | |
| "loss": 0.0215, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 1.0361445783132531e-05, | |
| "loss": 0.0446, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.0301204819277109e-05, | |
| "loss": 0.0491, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1.0240963855421688e-05, | |
| "loss": 0.0092, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1.0180722891566266e-05, | |
| "loss": 0.0031, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 1.0120481927710844e-05, | |
| "loss": 0.0076, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 1.0060240963855423e-05, | |
| "loss": 0.0642, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 1e-05, | |
| "loss": 0.029, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 9.93975903614458e-06, | |
| "loss": 0.0455, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 9.879518072289156e-06, | |
| "loss": 0.0097, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 9.819277108433736e-06, | |
| "loss": 0.0177, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 9.759036144578315e-06, | |
| "loss": 0.0047, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 9.698795180722893e-06, | |
| "loss": 0.0494, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 9.63855421686747e-06, | |
| "loss": 0.0237, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 9.57831325301205e-06, | |
| "loss": 0.0227, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 9.518072289156628e-06, | |
| "loss": 0.0109, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 9.457831325301205e-06, | |
| "loss": 0.0373, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 9.397590361445785e-06, | |
| "loss": 0.0439, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 9.337349397590362e-06, | |
| "loss": 0.0238, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 9.27710843373494e-06, | |
| "loss": 0.0174, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 9.21686746987952e-06, | |
| "loss": 0.0307, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 9.156626506024097e-06, | |
| "loss": 0.0078, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 9.096385542168675e-06, | |
| "loss": 0.021, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 9.036144578313254e-06, | |
| "loss": 0.0105, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 8.975903614457832e-06, | |
| "loss": 0.0495, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 8.91566265060241e-06, | |
| "loss": 0.064, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 8.855421686746989e-06, | |
| "loss": 0.063, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 8.795180722891567e-06, | |
| "loss": 0.0101, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 8.734939759036145e-06, | |
| "loss": 0.0184, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 8.674698795180724e-06, | |
| "loss": 0.0101, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 8.614457831325302e-06, | |
| "loss": 0.0083, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 8.55421686746988e-06, | |
| "loss": 0.0276, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 8.493975903614459e-06, | |
| "loss": 0.0099, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 8.433734939759038e-06, | |
| "loss": 0.0391, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 8.373493975903614e-06, | |
| "loss": 0.0512, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 8.313253012048194e-06, | |
| "loss": 0.0343, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 8.253012048192773e-06, | |
| "loss": 0.0273, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.19277108433735e-06, | |
| "loss": 0.0009, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 8.132530120481928e-06, | |
| "loss": 0.0356, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 8.072289156626508e-06, | |
| "loss": 0.0098, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 8.012048192771085e-06, | |
| "loss": 0.0311, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9976, | |
| "eval_loss": 0.011466823518276215, | |
| "eval_runtime": 67.0357, | |
| "eval_samples_per_second": 55.94, | |
| "eval_steps_per_second": 3.506, | |
| "step": 1992 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 7.951807228915663e-06, | |
| "loss": 0.0243, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 7.891566265060243e-06, | |
| "loss": 0.0081, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 7.83132530120482e-06, | |
| "loss": 0.0265, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.771084337349398e-06, | |
| "loss": 0.04, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 7.710843373493977e-06, | |
| "loss": 0.0384, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 7.650602409638555e-06, | |
| "loss": 0.0191, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 7.590361445783133e-06, | |
| "loss": 0.0169, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 7.530120481927712e-06, | |
| "loss": 0.0238, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 7.469879518072289e-06, | |
| "loss": 0.0189, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 7.4096385542168684e-06, | |
| "loss": 0.0101, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 7.349397590361447e-06, | |
| "loss": 0.0091, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 7.289156626506025e-06, | |
| "loss": 0.0504, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 7.228915662650603e-06, | |
| "loss": 0.0033, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 7.168674698795182e-06, | |
| "loss": 0.0103, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 7.1084337349397595e-06, | |
| "loss": 0.003, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 7.048192771084338e-06, | |
| "loss": 0.0376, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 6.987951807228917e-06, | |
| "loss": 0.0066, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 6.927710843373494e-06, | |
| "loss": 0.0293, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 6.867469879518073e-06, | |
| "loss": 0.0295, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 6.8072289156626514e-06, | |
| "loss": 0.0288, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 6.746987951807229e-06, | |
| "loss": 0.0095, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 6.686746987951808e-06, | |
| "loss": 0.0345, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 6.626506024096386e-06, | |
| "loss": 0.0362, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 6.566265060240964e-06, | |
| "loss": 0.0119, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 6.5060240963855425e-06, | |
| "loss": 0.0229, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 6.445783132530121e-06, | |
| "loss": 0.0568, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 6.385542168674699e-06, | |
| "loss": 0.0116, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 6.325301204819277e-06, | |
| "loss": 0.0053, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 6.265060240963856e-06, | |
| "loss": 0.0192, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 6.2048192771084344e-06, | |
| "loss": 0.0019, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 6.144578313253012e-06, | |
| "loss": 0.015, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 6.084337349397591e-06, | |
| "loss": 0.01, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 6.02409638554217e-06, | |
| "loss": 0.0764, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 5.963855421686747e-06, | |
| "loss": 0.0064, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 5.9036144578313255e-06, | |
| "loss": 0.0265, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 5.843373493975905e-06, | |
| "loss": 0.0166, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 5.783132530120482e-06, | |
| "loss": 0.0071, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 5.722891566265061e-06, | |
| "loss": 0.0306, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 5.66265060240964e-06, | |
| "loss": 0.0247, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 5.602409638554217e-06, | |
| "loss": 0.0178, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 5.542168674698796e-06, | |
| "loss": 0.0692, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 5.4819277108433745e-06, | |
| "loss": 0.0427, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 5.421686746987952e-06, | |
| "loss": 0.0457, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 5.361445783132531e-06, | |
| "loss": 0.027, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 5.301204819277109e-06, | |
| "loss": 0.0195, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 5.240963855421687e-06, | |
| "loss": 0.0173, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 5.180722891566266e-06, | |
| "loss": 0.0323, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 5.120481927710844e-06, | |
| "loss": 0.0642, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 5.060240963855422e-06, | |
| "loss": 0.0425, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0222, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 4.939759036144578e-06, | |
| "loss": 0.0122, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 4.8795180722891575e-06, | |
| "loss": 0.0217, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 4.819277108433735e-06, | |
| "loss": 0.0012, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 4.759036144578314e-06, | |
| "loss": 0.0696, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 4.698795180722892e-06, | |
| "loss": 0.0639, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 4.63855421686747e-06, | |
| "loss": 0.0033, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 4.578313253012049e-06, | |
| "loss": 0.0184, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 4.518072289156627e-06, | |
| "loss": 0.0189, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 4.457831325301205e-06, | |
| "loss": 0.0056, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 4.397590361445783e-06, | |
| "loss": 0.0347, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 4.337349397590362e-06, | |
| "loss": 0.0356, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 4.27710843373494e-06, | |
| "loss": 0.0775, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 4.216867469879519e-06, | |
| "loss": 0.0122, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 4.156626506024097e-06, | |
| "loss": 0.0035, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 4.096385542168675e-06, | |
| "loss": 0.0165, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 4.036144578313254e-06, | |
| "loss": 0.0021, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9973333333333333, | |
| "eval_loss": 0.012240640819072723, | |
| "eval_runtime": 67.2316, | |
| "eval_samples_per_second": 55.777, | |
| "eval_steps_per_second": 3.495, | |
| "step": 2656 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.975903614457832e-06, | |
| "loss": 0.0173, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.91566265060241e-06, | |
| "loss": 0.0375, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.855421686746989e-06, | |
| "loss": 0.0231, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.7951807228915664e-06, | |
| "loss": 0.0203, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.7349397590361445e-06, | |
| "loss": 0.0185, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.6746987951807235e-06, | |
| "loss": 0.0011, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.6144578313253016e-06, | |
| "loss": 0.0221, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 3.5542168674698798e-06, | |
| "loss": 0.0231, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.4939759036144583e-06, | |
| "loss": 0.0253, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.4337349397590364e-06, | |
| "loss": 0.0408, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 3.3734939759036146e-06, | |
| "loss": 0.049, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.313253012048193e-06, | |
| "loss": 0.0212, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.2530120481927713e-06, | |
| "loss": 0.0331, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.1927710843373494e-06, | |
| "loss": 0.0023, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.132530120481928e-06, | |
| "loss": 0.0166, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 3.072289156626506e-06, | |
| "loss": 0.0221, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.012048192771085e-06, | |
| "loss": 0.0155, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 2.9518072289156627e-06, | |
| "loss": 0.0218, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 2.891566265060241e-06, | |
| "loss": 0.0255, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 2.83132530120482e-06, | |
| "loss": 0.058, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 2.771084337349398e-06, | |
| "loss": 0.0362, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 2.710843373493976e-06, | |
| "loss": 0.067, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 2.6506024096385547e-06, | |
| "loss": 0.0081, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 2.590361445783133e-06, | |
| "loss": 0.0148, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 2.530120481927711e-06, | |
| "loss": 0.04, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 2.469879518072289e-06, | |
| "loss": 0.0005, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 2.4096385542168676e-06, | |
| "loss": 0.0206, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 2.349397590361446e-06, | |
| "loss": 0.0023, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 2.2891566265060243e-06, | |
| "loss": 0.0257, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 2.2289156626506024e-06, | |
| "loss": 0.0153, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.168674698795181e-06, | |
| "loss": 0.0218, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 2.1084337349397595e-06, | |
| "loss": 0.0154, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 2.0481927710843377e-06, | |
| "loss": 0.0231, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 1.987951807228916e-06, | |
| "loss": 0.0442, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 1.9277108433734943e-06, | |
| "loss": 0.0489, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 1.8674698795180723e-06, | |
| "loss": 0.0259, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 1.8072289156626508e-06, | |
| "loss": 0.0228, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 1.7469879518072292e-06, | |
| "loss": 0.0111, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 1.6867469879518073e-06, | |
| "loss": 0.0312, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 1.6265060240963856e-06, | |
| "loss": 0.0114, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 1.566265060240964e-06, | |
| "loss": 0.0066, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 1.5060240963855425e-06, | |
| "loss": 0.007, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 1.4457831325301204e-06, | |
| "loss": 0.0167, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 1.385542168674699e-06, | |
| "loss": 0.0005, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 1.3253012048192773e-06, | |
| "loss": 0.0177, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 1.2650602409638555e-06, | |
| "loss": 0.018, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 1.2048192771084338e-06, | |
| "loss": 0.0324, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 1.1445783132530121e-06, | |
| "loss": 0.0204, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 1.0843373493975905e-06, | |
| "loss": 0.0156, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 1.0240963855421688e-06, | |
| "loss": 0.021, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 9.638554216867472e-07, | |
| "loss": 0.0058, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 9.036144578313254e-07, | |
| "loss": 0.0201, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 8.433734939759036e-07, | |
| "loss": 0.0566, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 7.83132530120482e-07, | |
| "loss": 0.0018, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 7.228915662650602e-07, | |
| "loss": 0.0205, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 6.626506024096387e-07, | |
| "loss": 0.0341, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 6.024096385542169e-07, | |
| "loss": 0.0102, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 5.421686746987952e-07, | |
| "loss": 0.006, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 4.819277108433736e-07, | |
| "loss": 0.0252, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 4.216867469879518e-07, | |
| "loss": 0.0373, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 3.614457831325301e-07, | |
| "loss": 0.011, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 3.0120481927710845e-07, | |
| "loss": 0.009, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 2.409638554216868e-07, | |
| "loss": 0.0332, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 1.8072289156626505e-07, | |
| "loss": 0.0596, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 1.204819277108434e-07, | |
| "loss": 0.0162, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 6.02409638554217e-08, | |
| "loss": 0.0419, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.0248, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9973333333333333, | |
| "eval_loss": 0.010264705866575241, | |
| "eval_runtime": 67.0312, | |
| "eval_samples_per_second": 55.944, | |
| "eval_steps_per_second": 3.506, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 3320, | |
| "total_flos": 8.402423822265876e+18, | |
| "train_loss": 0.031068402516560248, | |
| "train_runtime": 5109.0232, | |
| "train_samples_per_second": 20.795, | |
| "train_steps_per_second": 0.65 | |
| } | |
| ], | |
| "max_steps": 3320, | |
| "num_train_epochs": 5, | |
| "total_flos": 8.402423822265876e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |