convnextv2-large-1k-224-finetuned-LungCancer-Classification-LC25000-AH-60-20-20
/
trainer_state.json
| { | |
| "best_metric": 0.9996666666666667, | |
| "best_model_checkpoint": "convnextv2-large-1k-224-finetuned-LungCancer-Classification-LC25000-AH-60-20-20/checkpoint-2252", | |
| "epoch": 11.978685612788633, | |
| "global_step": 3372, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 8.237232289950577e-08, | |
| "loss": 1.1137, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.6474464579901155e-07, | |
| "loss": 1.1038, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.471169686985173e-07, | |
| "loss": 1.1058, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.294892915980231e-07, | |
| "loss": 1.1019, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.1186161449752885e-07, | |
| "loss": 1.0968, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.942339373970346e-07, | |
| "loss": 1.1, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 5.766062602965404e-07, | |
| "loss": 1.0923, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 6.589785831960462e-07, | |
| "loss": 1.0896, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.413509060955519e-07, | |
| "loss": 1.0866, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.237232289950577e-07, | |
| "loss": 1.0871, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.060955518945634e-07, | |
| "loss": 1.0814, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.884678747940691e-07, | |
| "loss": 1.0853, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.0708401976935751e-06, | |
| "loss": 1.0734, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.1532125205930807e-06, | |
| "loss": 1.0779, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.2355848434925866e-06, | |
| "loss": 1.0607, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.3179571663920924e-06, | |
| "loss": 1.0505, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.4003294892915982e-06, | |
| "loss": 1.0551, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.4827018121911038e-06, | |
| "loss": 1.038, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.5650741350906096e-06, | |
| "loss": 1.029, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.6474464579901154e-06, | |
| "loss": 1.0264, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.729818780889621e-06, | |
| "loss": 1.0118, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.8121911037891268e-06, | |
| "loss": 1.007, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.8945634266886326e-06, | |
| "loss": 0.9894, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.9769357495881382e-06, | |
| "loss": 0.9815, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.059308072487644e-06, | |
| "loss": 0.971, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.1416803953871503e-06, | |
| "loss": 0.9631, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.224052718286656e-06, | |
| "loss": 0.9367, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.3064250411861615e-06, | |
| "loss": 0.928, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.3887973640856673e-06, | |
| "loss": 0.9194, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.471169686985173e-06, | |
| "loss": 0.8981, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.553542009884679e-06, | |
| "loss": 0.893, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.6359143327841847e-06, | |
| "loss": 0.87, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.7182866556836905e-06, | |
| "loss": 0.8457, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.8006589785831964e-06, | |
| "loss": 0.8226, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 2.8830313014827017e-06, | |
| "loss": 0.8054, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.9654036243822076e-06, | |
| "loss": 0.8215, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.0477759472817134e-06, | |
| "loss": 0.7601, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.130148270181219e-06, | |
| "loss": 0.7473, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.212520593080725e-06, | |
| "loss": 0.7349, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.294892915980231e-06, | |
| "loss": 0.7256, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.3772652388797366e-06, | |
| "loss": 0.7132, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.459637561779242e-06, | |
| "loss": 0.6912, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.542009884678748e-06, | |
| "loss": 0.6832, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.6243822075782536e-06, | |
| "loss": 0.6468, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.7067545304777594e-06, | |
| "loss": 0.6275, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.7891268533772653e-06, | |
| "loss": 0.6068, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.871499176276771e-06, | |
| "loss": 0.5861, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.9538714991762765e-06, | |
| "loss": 0.5385, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.036243822075783e-06, | |
| "loss": 0.5482, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.118616144975288e-06, | |
| "loss": 0.4988, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.200988467874794e-06, | |
| "loss": 0.4763, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.2833607907743006e-06, | |
| "loss": 0.4613, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.365733113673806e-06, | |
| "loss": 0.4527, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.448105436573312e-06, | |
| "loss": 0.4072, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.5304777594728176e-06, | |
| "loss": 0.4084, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.612850082372323e-06, | |
| "loss": 0.4332, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.947, | |
| "eval_loss": 0.3823847770690918, | |
| "eval_runtime": 102.7646, | |
| "eval_samples_per_second": 29.193, | |
| "eval_steps_per_second": 1.829, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.695222405271829e-06, | |
| "loss": 0.3971, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.777594728171335e-06, | |
| "loss": 0.4011, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.859967051070841e-06, | |
| "loss": 0.3385, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.942339373970346e-06, | |
| "loss": 0.3713, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 5.0247116968698524e-06, | |
| "loss": 0.3593, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 5.107084019769358e-06, | |
| "loss": 0.3169, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.189456342668863e-06, | |
| "loss": 0.3527, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.2718286655683695e-06, | |
| "loss": 0.3359, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.354200988467875e-06, | |
| "loss": 0.3294, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.436573311367381e-06, | |
| "loss": 0.367, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.5189456342668865e-06, | |
| "loss": 0.2967, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.601317957166393e-06, | |
| "loss": 0.3344, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.683690280065898e-06, | |
| "loss": 0.2708, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 5.7660626029654035e-06, | |
| "loss": 0.2952, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.84843492586491e-06, | |
| "loss": 0.3169, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.930807248764415e-06, | |
| "loss": 0.2502, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 6.013179571663921e-06, | |
| "loss": 0.2483, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.095551894563427e-06, | |
| "loss": 0.2372, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 6.177924217462933e-06, | |
| "loss": 0.273, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 6.260296540362438e-06, | |
| "loss": 0.2797, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 6.342668863261944e-06, | |
| "loss": 0.2496, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 6.42504118616145e-06, | |
| "loss": 0.2831, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 6.507413509060955e-06, | |
| "loss": 0.2257, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 6.589785831960462e-06, | |
| "loss": 0.2512, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 6.672158154859967e-06, | |
| "loss": 0.2414, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 6.754530477759473e-06, | |
| "loss": 0.2722, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.836902800658979e-06, | |
| "loss": 0.236, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.919275123558484e-06, | |
| "loss": 0.2006, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.00164744645799e-06, | |
| "loss": 0.2725, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.084019769357496e-06, | |
| "loss": 0.22, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.166392092257002e-06, | |
| "loss": 0.2074, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 7.248764415156507e-06, | |
| "loss": 0.1888, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 7.3311367380560135e-06, | |
| "loss": 0.2646, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 7.413509060955519e-06, | |
| "loss": 0.2729, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 7.495881383855024e-06, | |
| "loss": 0.2633, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 7.5782537067545305e-06, | |
| "loss": 0.2753, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 7.660626029654036e-06, | |
| "loss": 0.2074, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 7.742998352553542e-06, | |
| "loss": 0.165, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 7.825370675453048e-06, | |
| "loss": 0.2505, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 7.907742998352553e-06, | |
| "loss": 0.2062, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 7.99011532125206e-06, | |
| "loss": 0.2255, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.072487644151565e-06, | |
| "loss": 0.2634, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 8.15485996705107e-06, | |
| "loss": 0.2253, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.237232289950576e-06, | |
| "loss": 0.1968, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.319604612850082e-06, | |
| "loss": 0.1252, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 8.401976935749589e-06, | |
| "loss": 0.152, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 8.484349258649095e-06, | |
| "loss": 0.154, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 8.566721581548601e-06, | |
| "loss": 0.2983, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 8.649093904448106e-06, | |
| "loss": 0.2407, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 8.731466227347612e-06, | |
| "loss": 0.3075, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.813838550247118e-06, | |
| "loss": 0.1653, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.896210873146624e-06, | |
| "loss": 0.1378, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.978583196046129e-06, | |
| "loss": 0.1574, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 9.060955518945635e-06, | |
| "loss": 0.1756, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 9.143327841845141e-06, | |
| "loss": 0.3551, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 9.225700164744646e-06, | |
| "loss": 0.1558, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9736666666666667, | |
| "eval_loss": 0.12916548550128937, | |
| "eval_runtime": 102.8832, | |
| "eval_samples_per_second": 29.159, | |
| "eval_steps_per_second": 1.827, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.308072487644152e-06, | |
| "loss": 0.2096, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.390444810543658e-06, | |
| "loss": 0.1701, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.472817133443165e-06, | |
| "loss": 0.1898, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 9.55518945634267e-06, | |
| "loss": 0.1353, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.637561779242175e-06, | |
| "loss": 0.2444, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.719934102141682e-06, | |
| "loss": 0.1267, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 9.802306425041186e-06, | |
| "loss": 0.1643, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.884678747940692e-06, | |
| "loss": 0.1876, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 9.967051070840199e-06, | |
| "loss": 0.1381, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.0049423393739705e-05, | |
| "loss": 0.1257, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.013179571663921e-05, | |
| "loss": 0.1204, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.0214168039538716e-05, | |
| "loss": 0.134, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.0296540362438222e-05, | |
| "loss": 0.1183, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.0378912685337726e-05, | |
| "loss": 0.1981, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.0461285008237233e-05, | |
| "loss": 0.0772, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.0543657331136739e-05, | |
| "loss": 0.1575, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.0626029654036245e-05, | |
| "loss": 0.1043, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.070840197693575e-05, | |
| "loss": 0.162, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.0790774299835256e-05, | |
| "loss": 0.1435, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.0873146622734762e-05, | |
| "loss": 0.1237, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0955518945634267e-05, | |
| "loss": 0.1999, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.1037891268533773e-05, | |
| "loss": 0.1393, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.112026359143328e-05, | |
| "loss": 0.1782, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.1202635914332785e-05, | |
| "loss": 0.1328, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 1.128500823723229e-05, | |
| "loss": 0.1462, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 1.1367380560131796e-05, | |
| "loss": 0.3144, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 1.1449752883031302e-05, | |
| "loss": 0.0946, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 1.1532125205930807e-05, | |
| "loss": 0.1079, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 1.1614497528830313e-05, | |
| "loss": 0.1672, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 1.169686985172982e-05, | |
| "loss": 0.0851, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 1.1779242174629326e-05, | |
| "loss": 0.08, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1.186161449752883e-05, | |
| "loss": 0.0899, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1.1943986820428336e-05, | |
| "loss": 0.0875, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 1.2026359143327843e-05, | |
| "loss": 0.1972, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 1.2108731466227347e-05, | |
| "loss": 0.2168, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 1.2191103789126853e-05, | |
| "loss": 0.119, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 1.227347611202636e-05, | |
| "loss": 0.2255, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 1.2355848434925866e-05, | |
| "loss": 0.0626, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 1.243822075782537e-05, | |
| "loss": 0.0585, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 1.2520593080724877e-05, | |
| "loss": 0.2077, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.2602965403624385e-05, | |
| "loss": 0.0992, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.2685337726523888e-05, | |
| "loss": 0.1492, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.2767710049423395e-05, | |
| "loss": 0.186, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.28500823723229e-05, | |
| "loss": 0.1707, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.2932454695222406e-05, | |
| "loss": 0.0914, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.301482701812191e-05, | |
| "loss": 0.097, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.3097199341021419e-05, | |
| "loss": 0.2155, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.3179571663920923e-05, | |
| "loss": 0.1728, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 1.326194398682043e-05, | |
| "loss": 0.1858, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1.3344316309719934e-05, | |
| "loss": 0.2391, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.3426688632619442e-05, | |
| "loss": 0.1996, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.3509060955518946e-05, | |
| "loss": 0.164, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.3591433278418453e-05, | |
| "loss": 0.1101, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 1.3673805601317957e-05, | |
| "loss": 0.1583, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 1.3756177924217463e-05, | |
| "loss": 0.061, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.3838550247116968e-05, | |
| "loss": 0.1161, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9886666666666667, | |
| "eval_loss": 0.05563023313879967, | |
| "eval_runtime": 102.7778, | |
| "eval_samples_per_second": 29.189, | |
| "eval_steps_per_second": 1.829, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1.3920922570016476e-05, | |
| "loss": 0.1593, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.400329489291598e-05, | |
| "loss": 0.1887, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.4085667215815487e-05, | |
| "loss": 0.147, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.4168039538714991e-05, | |
| "loss": 0.1935, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 1.42504118616145e-05, | |
| "loss": 0.0872, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 1.4332784184514004e-05, | |
| "loss": 0.1438, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.441515650741351e-05, | |
| "loss": 0.1659, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 1.4497528830313015e-05, | |
| "loss": 0.1528, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 1.4579901153212522e-05, | |
| "loss": 0.1848, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 1.4662273476112027e-05, | |
| "loss": 0.113, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.4744645799011533e-05, | |
| "loss": 0.1683, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.4827018121911038e-05, | |
| "loss": 0.0835, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.4909390444810544e-05, | |
| "loss": 0.0722, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.4991762767710049e-05, | |
| "loss": 0.0952, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.5074135090609557e-05, | |
| "loss": 0.0605, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.5156507413509061e-05, | |
| "loss": 0.029, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.5238879736408567e-05, | |
| "loss": 0.1104, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 1.5321252059308072e-05, | |
| "loss": 0.1488, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.540362438220758e-05, | |
| "loss": 0.1516, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 1.5485996705107084e-05, | |
| "loss": 0.1192, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.5568369028006592e-05, | |
| "loss": 0.2367, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 1.5650741350906097e-05, | |
| "loss": 0.1001, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.57331136738056e-05, | |
| "loss": 0.1122, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.5815485996705106e-05, | |
| "loss": 0.1992, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.5897858319604614e-05, | |
| "loss": 0.0839, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 1.598023064250412e-05, | |
| "loss": 0.1391, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 1.6062602965403626e-05, | |
| "loss": 0.1759, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 1.614497528830313e-05, | |
| "loss": 0.0337, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 1.6227347611202635e-05, | |
| "loss": 0.1444, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 1.630971993410214e-05, | |
| "loss": 0.0905, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 1.6392092257001648e-05, | |
| "loss": 0.1339, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.6474464579901152e-05, | |
| "loss": 0.0834, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 1.655683690280066e-05, | |
| "loss": 0.0958, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 1.6639209225700165e-05, | |
| "loss": 0.054, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 1.6721581548599673e-05, | |
| "loss": 0.1113, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 1.6803953871499177e-05, | |
| "loss": 0.1611, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 1.6886326194398682e-05, | |
| "loss": 0.1001, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 1.696869851729819e-05, | |
| "loss": 0.0998, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 1.7051070840197694e-05, | |
| "loss": 0.0603, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 1.7133443163097202e-05, | |
| "loss": 0.1444, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 1.7215815485996707e-05, | |
| "loss": 0.1594, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 1.729818780889621e-05, | |
| "loss": 0.1355, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 1.7380560131795716e-05, | |
| "loss": 0.1373, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.7462932454695224e-05, | |
| "loss": 0.0969, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.754530477759473e-05, | |
| "loss": 0.0711, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 1.7627677100494236e-05, | |
| "loss": 0.1425, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 1.771004942339374e-05, | |
| "loss": 0.1354, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 1.779242174629325e-05, | |
| "loss": 0.183, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 1.7874794069192753e-05, | |
| "loss": 0.0864, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 1.7957166392092258e-05, | |
| "loss": 0.11, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.8039538714991762e-05, | |
| "loss": 0.0572, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 1.812191103789127e-05, | |
| "loss": 0.0901, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.8204283360790775e-05, | |
| "loss": 0.0958, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.8286655683690283e-05, | |
| "loss": 0.1022, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 1.8369028006589787e-05, | |
| "loss": 0.0814, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.8451400329489292e-05, | |
| "loss": 0.0505, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.8533772652388796e-05, | |
| "loss": 0.2337, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.982, | |
| "eval_loss": 0.06833887100219727, | |
| "eval_runtime": 103.3312, | |
| "eval_samples_per_second": 29.033, | |
| "eval_steps_per_second": 1.819, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 1.8616144975288304e-05, | |
| "loss": 0.1804, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.869851729818781e-05, | |
| "loss": 0.0369, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 1.8780889621087317e-05, | |
| "loss": 0.1201, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.886326194398682e-05, | |
| "loss": 0.1551, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.894563426688633e-05, | |
| "loss": 0.0762, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.9028006589785834e-05, | |
| "loss": 0.0775, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 1.911037891268534e-05, | |
| "loss": 0.0529, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 1.9192751235584843e-05, | |
| "loss": 0.0614, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 1.927512355848435e-05, | |
| "loss": 0.0438, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 1.9357495881383855e-05, | |
| "loss": 0.0592, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 1.9439868204283363e-05, | |
| "loss": 0.0795, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 1.9522240527182868e-05, | |
| "loss": 0.154, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 1.9604612850082372e-05, | |
| "loss": 0.1057, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 1.9686985172981877e-05, | |
| "loss": 0.0702, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 1.9769357495881385e-05, | |
| "loss": 0.1535, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 1.985172981878089e-05, | |
| "loss": 0.125, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 1.9934102141680397e-05, | |
| "loss": 0.1554, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 2.0016474464579902e-05, | |
| "loss": 0.2754, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 2.009884678747941e-05, | |
| "loss": 0.1334, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 2.0181219110378914e-05, | |
| "loss": 0.1489, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 2.026359143327842e-05, | |
| "loss": 0.125, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 2.0345963756177923e-05, | |
| "loss": 0.0825, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 2.042833607907743e-05, | |
| "loss": 0.0359, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 2.0510708401976936e-05, | |
| "loss": 0.1447, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 2.0593080724876444e-05, | |
| "loss": 0.0958, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.067545304777595e-05, | |
| "loss": 0.1453, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 2.0757825370675453e-05, | |
| "loss": 0.0814, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 2.0840197693574957e-05, | |
| "loss": 0.1224, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 2.0922570016474465e-05, | |
| "loss": 0.0573, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 2.100494233937397e-05, | |
| "loss": 0.06, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 2.1087314662273478e-05, | |
| "loss": 0.0671, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 2.1169686985172982e-05, | |
| "loss": 0.0732, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 2.125205930807249e-05, | |
| "loss": 0.0669, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 2.1334431630971995e-05, | |
| "loss": 0.0719, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 2.14168039538715e-05, | |
| "loss": 0.0729, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 2.1499176276771004e-05, | |
| "loss": 0.0659, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 2.1581548599670512e-05, | |
| "loss": 0.0926, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.1663920922570016e-05, | |
| "loss": 0.0682, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 2.1746293245469524e-05, | |
| "loss": 0.1173, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 2.182866556836903e-05, | |
| "loss": 0.1075, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.1911037891268533e-05, | |
| "loss": 0.0801, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.1993410214168038e-05, | |
| "loss": 0.1199, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.2075782537067546e-05, | |
| "loss": 0.0687, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 2.215815485996705e-05, | |
| "loss": 0.0116, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 2.224052718286656e-05, | |
| "loss": 0.0827, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 2.2322899505766066e-05, | |
| "loss": 0.1503, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.240527182866557e-05, | |
| "loss": 0.1168, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 2.2487644151565075e-05, | |
| "loss": 0.0839, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 2.257001647446458e-05, | |
| "loss": 0.079, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 2.2652388797364088e-05, | |
| "loss": 0.0179, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 2.2734761120263592e-05, | |
| "loss": 0.1077, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 2.28171334431631e-05, | |
| "loss": 0.0753, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 2.2899505766062605e-05, | |
| "loss": 0.0599, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 2.298187808896211e-05, | |
| "loss": 0.1009, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 2.3064250411861614e-05, | |
| "loss": 0.1012, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 2.3146622734761122e-05, | |
| "loss": 0.1285, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.9923333333333333, | |
| "eval_loss": 0.02931647002696991, | |
| "eval_runtime": 102.8839, | |
| "eval_samples_per_second": 29.159, | |
| "eval_steps_per_second": 1.827, | |
| "step": 1407 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 2.3228995057660626e-05, | |
| "loss": 0.0156, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.3311367380560134e-05, | |
| "loss": 0.0601, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 2.339373970345964e-05, | |
| "loss": 0.1066, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 2.3476112026359147e-05, | |
| "loss": 0.1083, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.355848434925865e-05, | |
| "loss": 0.1596, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 2.3640856672158156e-05, | |
| "loss": 0.0789, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.372322899505766e-05, | |
| "loss": 0.1064, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 2.380560131795717e-05, | |
| "loss": 0.1128, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 2.3887973640856673e-05, | |
| "loss": 0.1064, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 2.397034596375618e-05, | |
| "loss": 0.0708, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 2.4052718286655685e-05, | |
| "loss": 0.1121, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 2.413509060955519e-05, | |
| "loss": 0.0615, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 2.4217462932454694e-05, | |
| "loss": 0.2011, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 2.4299835255354202e-05, | |
| "loss": 0.0912, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 2.4382207578253707e-05, | |
| "loss": 0.0729, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 2.4464579901153215e-05, | |
| "loss": 0.0371, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 2.454695222405272e-05, | |
| "loss": 0.0649, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 2.4629324546952227e-05, | |
| "loss": 0.0957, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 2.4711696869851732e-05, | |
| "loss": 0.1118, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 2.4794069192751236e-05, | |
| "loss": 0.0885, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 2.487644151565074e-05, | |
| "loss": 0.0483, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 2.495881383855025e-05, | |
| "loss": 0.0613, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 2.5041186161449753e-05, | |
| "loss": 0.1008, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 2.5123558484349258e-05, | |
| "loss": 0.1557, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 2.520593080724877e-05, | |
| "loss": 0.0303, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 2.528830313014827e-05, | |
| "loss": 0.0462, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 2.5370675453047775e-05, | |
| "loss": 0.0925, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 2.545304777594728e-05, | |
| "loss": 0.061, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 2.553542009884679e-05, | |
| "loss": 0.0585, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 2.5617792421746295e-05, | |
| "loss": 0.0655, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 2.57001647446458e-05, | |
| "loss": 0.1573, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 2.5782537067545305e-05, | |
| "loss": 0.0872, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 2.5864909390444812e-05, | |
| "loss": 0.0605, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 2.5947281713344317e-05, | |
| "loss": 0.0434, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 2.602965403624382e-05, | |
| "loss": 0.0977, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 2.6112026359143326e-05, | |
| "loss": 0.0319, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 2.6194398682042837e-05, | |
| "loss": 0.0557, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 2.6276771004942342e-05, | |
| "loss": 0.1051, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 2.6359143327841846e-05, | |
| "loss": 0.0481, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 2.644151565074135e-05, | |
| "loss": 0.0303, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 2.652388797364086e-05, | |
| "loss": 0.1231, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 2.6606260296540363e-05, | |
| "loss": 0.044, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 2.6688632619439868e-05, | |
| "loss": 0.0462, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 2.6771004942339373e-05, | |
| "loss": 0.036, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 2.6853377265238884e-05, | |
| "loss": 0.0388, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 2.693574958813839e-05, | |
| "loss": 0.1279, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 2.7018121911037893e-05, | |
| "loss": 0.0419, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 2.7100494233937394e-05, | |
| "loss": 0.1365, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 2.7182866556836905e-05, | |
| "loss": 0.0455, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 2.726523887973641e-05, | |
| "loss": 0.1434, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 2.7347611202635915e-05, | |
| "loss": 0.0875, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 2.742998352553542e-05, | |
| "loss": 0.1429, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 2.7512355848434927e-05, | |
| "loss": 0.0456, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 2.759472817133443e-05, | |
| "loss": 0.1056, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 2.7677100494233936e-05, | |
| "loss": 0.0315, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 2.775947281713344e-05, | |
| "loss": 0.047, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.975, | |
| "eval_loss": 0.09873583912849426, | |
| "eval_runtime": 103.1743, | |
| "eval_samples_per_second": 29.077, | |
| "eval_steps_per_second": 1.822, | |
| "step": 1689 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2.7841845140032952e-05, | |
| "loss": 0.0723, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.7924217462932456e-05, | |
| "loss": 0.287, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 2.800658978583196e-05, | |
| "loss": 0.0364, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.808896210873147e-05, | |
| "loss": 0.1763, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 2.8171334431630973e-05, | |
| "loss": 0.1123, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 2.8253706754530478e-05, | |
| "loss": 0.1386, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 2.8336079077429983e-05, | |
| "loss": 0.0935, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 2.8418451400329494e-05, | |
| "loss": 0.0552, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 2.8500823723229e-05, | |
| "loss": 0.1309, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 2.8583196046128503e-05, | |
| "loss": 0.0556, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 2.8665568369028008e-05, | |
| "loss": 0.0378, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 2.8747940691927515e-05, | |
| "loss": 0.0699, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 2.883031301482702e-05, | |
| "loss": 0.0824, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 2.8912685337726525e-05, | |
| "loss": 0.1064, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 2.899505766062603e-05, | |
| "loss": 0.0639, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 2.907742998352554e-05, | |
| "loss": 0.0946, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 2.9159802306425045e-05, | |
| "loss": 0.0716, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "learning_rate": 2.924217462932455e-05, | |
| "loss": 0.0552, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 2.9324546952224054e-05, | |
| "loss": 0.0617, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 2.9406919275123562e-05, | |
| "loss": 0.0163, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 2.9489291598023067e-05, | |
| "loss": 0.0336, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 2.957166392092257e-05, | |
| "loss": 0.0533, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 2.9654036243822076e-05, | |
| "loss": 0.0402, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 2.9736408566721587e-05, | |
| "loss": 0.0353, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 2.9818780889621088e-05, | |
| "loss": 0.0737, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 2.9901153212520593e-05, | |
| "loss": 0.1074, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 2.9983525535420097e-05, | |
| "loss": 0.0236, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 3.006589785831961e-05, | |
| "loss": 0.075, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 3.0148270181219113e-05, | |
| "loss": 0.0617, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 3.0230642504118618e-05, | |
| "loss": 0.0139, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 3.0313014827018122e-05, | |
| "loss": 0.028, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "learning_rate": 3.039538714991763e-05, | |
| "loss": 0.0746, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 3.0477759472817135e-05, | |
| "loss": 0.1602, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 3.056013179571664e-05, | |
| "loss": 0.0461, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "learning_rate": 3.0642504118616144e-05, | |
| "loss": 0.0324, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 3.0724876441515655e-05, | |
| "loss": 0.0803, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 3.080724876441516e-05, | |
| "loss": 0.073, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 3.0889621087314664e-05, | |
| "loss": 0.1086, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 3.097199341021417e-05, | |
| "loss": 0.0582, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 3.105436573311368e-05, | |
| "loss": 0.042, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 3.1136738056013184e-05, | |
| "loss": 0.1111, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "learning_rate": 3.121911037891269e-05, | |
| "loss": 0.0646, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 3.1301482701812194e-05, | |
| "loss": 0.1404, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 3.13838550247117e-05, | |
| "loss": 0.0556, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 3.14662273476112e-05, | |
| "loss": 0.1132, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 3.154859967051071e-05, | |
| "loss": 0.0547, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 3.163097199341021e-05, | |
| "loss": 0.0944, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 3.171334431630972e-05, | |
| "loss": 0.0772, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 3.179571663920923e-05, | |
| "loss": 0.1075, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "learning_rate": 3.187808896210873e-05, | |
| "loss": 0.2451, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 3.196046128500824e-05, | |
| "loss": 0.0491, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 3.204283360790775e-05, | |
| "loss": 0.1249, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 3.212520593080725e-05, | |
| "loss": 0.0504, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 6.94, | |
| "learning_rate": 3.220757825370676e-05, | |
| "loss": 0.0896, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 3.228995057660626e-05, | |
| "loss": 0.1948, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "learning_rate": 3.2372322899505766e-05, | |
| "loss": 0.1477, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3.245469522240527e-05, | |
| "loss": 0.0741, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.988, | |
| "eval_loss": 0.037325419485569, | |
| "eval_runtime": 103.071, | |
| "eval_samples_per_second": 29.106, | |
| "eval_steps_per_second": 1.824, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 3.2537067545304775e-05, | |
| "loss": 0.0566, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 3.261943986820428e-05, | |
| "loss": 0.0324, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 3.270181219110379e-05, | |
| "loss": 0.0586, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 3.2784184514003296e-05, | |
| "loss": 0.0519, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 3.28665568369028e-05, | |
| "loss": 0.0255, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 3.2948929159802305e-05, | |
| "loss": 0.0865, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 3.3031301482701816e-05, | |
| "loss": 0.0349, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 3.311367380560132e-05, | |
| "loss": 0.0897, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 3.3196046128500825e-05, | |
| "loss": 0.1209, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 3.327841845140033e-05, | |
| "loss": 0.0204, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "learning_rate": 3.336079077429984e-05, | |
| "loss": 0.1516, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 3.3443163097199346e-05, | |
| "loss": 0.0491, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 3.352553542009885e-05, | |
| "loss": 0.1446, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 3.3607907742998355e-05, | |
| "loss": 0.0114, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 3.369028006589786e-05, | |
| "loss": 0.0629, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 3.3772652388797364e-05, | |
| "loss": 0.0072, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 3.385502471169687e-05, | |
| "loss": 0.0152, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 3.393739703459638e-05, | |
| "loss": 0.1723, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 3.4019769357495884e-05, | |
| "loss": 0.0597, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 3.410214168039539e-05, | |
| "loss": 0.1124, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 3.418451400329489e-05, | |
| "loss": 0.066, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 3.4266886326194404e-05, | |
| "loss": 0.0129, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 3.434925864909391e-05, | |
| "loss": 0.0213, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 3.4431630971993414e-05, | |
| "loss": 0.189, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 7.44, | |
| "learning_rate": 3.451400329489292e-05, | |
| "loss": 0.1282, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 3.459637561779242e-05, | |
| "loss": 0.1049, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 3.467874794069193e-05, | |
| "loss": 0.0942, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 3.476112026359143e-05, | |
| "loss": 0.1089, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "learning_rate": 3.4843492586490936e-05, | |
| "loss": 0.0919, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 3.492586490939045e-05, | |
| "loss": 0.0271, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 3.500823723228995e-05, | |
| "loss": 0.0784, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 3.509060955518946e-05, | |
| "loss": 0.0201, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 3.517298187808896e-05, | |
| "loss": 0.0949, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 3.525535420098847e-05, | |
| "loss": 0.017, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 3.533772652388798e-05, | |
| "loss": 0.0507, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "learning_rate": 3.542009884678748e-05, | |
| "loss": 0.0248, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 3.5502471169686986e-05, | |
| "loss": 0.0257, | |
| "step": 2155 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 3.55848434925865e-05, | |
| "loss": 0.0722, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 3.5667215815486e-05, | |
| "loss": 0.0443, | |
| "step": 2165 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 3.5749588138385507e-05, | |
| "loss": 0.0622, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 3.583196046128501e-05, | |
| "loss": 0.0416, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 3.5914332784184516e-05, | |
| "loss": 0.0351, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 7.76, | |
| "learning_rate": 3.599670510708402e-05, | |
| "loss": 0.0827, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 3.6079077429983525e-05, | |
| "loss": 0.114, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 3.616144975288303e-05, | |
| "loss": 0.0579, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 3.624382207578254e-05, | |
| "loss": 0.0879, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 3.6326194398682045e-05, | |
| "loss": 0.0622, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 3.640856672158155e-05, | |
| "loss": 0.0688, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 3.6490939044481054e-05, | |
| "loss": 0.0285, | |
| "step": 2215 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 3.6573311367380566e-05, | |
| "loss": 0.1117, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 3.665568369028007e-05, | |
| "loss": 0.142, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 3.6738056013179575e-05, | |
| "loss": 0.0554, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 3.682042833607908e-05, | |
| "loss": 0.0532, | |
| "step": 2235 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 3.6902800658978584e-05, | |
| "loss": 0.0602, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 7.98, | |
| "learning_rate": 3.698517298187809e-05, | |
| "loss": 0.0585, | |
| "step": 2245 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 3.706754530477759e-05, | |
| "loss": 0.153, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.9996666666666667, | |
| "eval_loss": 0.004267791751772165, | |
| "eval_runtime": 103.6449, | |
| "eval_samples_per_second": 28.945, | |
| "eval_steps_per_second": 1.814, | |
| "step": 2252 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 3.71499176276771e-05, | |
| "loss": 0.0074, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 3.723228995057661e-05, | |
| "loss": 0.0601, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 3.731466227347611e-05, | |
| "loss": 0.1, | |
| "step": 2265 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 3.739703459637562e-05, | |
| "loss": 0.0091, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 3.747940691927512e-05, | |
| "loss": 0.1053, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 3.7561779242174634e-05, | |
| "loss": 0.0764, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 3.764415156507414e-05, | |
| "loss": 0.0205, | |
| "step": 2285 | |
| }, | |
| { | |
| "epoch": 8.13, | |
| "learning_rate": 3.772652388797364e-05, | |
| "loss": 0.0291, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 3.780889621087315e-05, | |
| "loss": 0.036, | |
| "step": 2295 | |
| }, | |
| { | |
| "epoch": 8.17, | |
| "learning_rate": 3.789126853377266e-05, | |
| "loss": 0.0944, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 3.797364085667216e-05, | |
| "loss": 0.035, | |
| "step": 2305 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 3.805601317957167e-05, | |
| "loss": 0.0377, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 3.813838550247117e-05, | |
| "loss": 0.07, | |
| "step": 2315 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 3.822075782537068e-05, | |
| "loss": 0.0173, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 3.830313014827018e-05, | |
| "loss": 0.1078, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 8.28, | |
| "learning_rate": 3.8385502471169686e-05, | |
| "loss": 0.0844, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 3.846787479406919e-05, | |
| "loss": 0.0506, | |
| "step": 2335 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 3.85502471169687e-05, | |
| "loss": 0.008, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 3.8632619439868206e-05, | |
| "loss": 0.0716, | |
| "step": 2345 | |
| }, | |
| { | |
| "epoch": 8.35, | |
| "learning_rate": 3.871499176276771e-05, | |
| "loss": 0.0405, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 3.8797364085667215e-05, | |
| "loss": 0.0507, | |
| "step": 2355 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 3.8879736408566727e-05, | |
| "loss": 0.0328, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 3.896210873146623e-05, | |
| "loss": 0.1797, | |
| "step": 2365 | |
| }, | |
| { | |
| "epoch": 8.42, | |
| "learning_rate": 3.9044481054365736e-05, | |
| "loss": 0.059, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 3.912685337726524e-05, | |
| "loss": 0.0714, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 8.45, | |
| "learning_rate": 3.9209225700164745e-05, | |
| "loss": 0.0615, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 3.929159802306425e-05, | |
| "loss": 0.0667, | |
| "step": 2385 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 3.9373970345963754e-05, | |
| "loss": 0.154, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 8.51, | |
| "learning_rate": 3.9456342668863265e-05, | |
| "loss": 0.0138, | |
| "step": 2395 | |
| }, | |
| { | |
| "epoch": 8.53, | |
| "learning_rate": 3.953871499176277e-05, | |
| "loss": 0.0342, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 3.9621087314662274e-05, | |
| "loss": 0.0212, | |
| "step": 2405 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 3.970345963756178e-05, | |
| "loss": 0.0792, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 3.978583196046129e-05, | |
| "loss": 0.1257, | |
| "step": 2415 | |
| }, | |
| { | |
| "epoch": 8.6, | |
| "learning_rate": 3.9868204283360795e-05, | |
| "loss": 0.1654, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 3.99505766062603e-05, | |
| "loss": 0.0687, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 4.0032948929159804e-05, | |
| "loss": 0.0887, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 8.65, | |
| "learning_rate": 4.0115321252059315e-05, | |
| "loss": 0.0506, | |
| "step": 2435 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 4.019769357495882e-05, | |
| "loss": 0.0882, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 4.0280065897858324e-05, | |
| "loss": 0.0695, | |
| "step": 2445 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 4.036243822075783e-05, | |
| "loss": 0.0082, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 4.044481054365733e-05, | |
| "loss": 0.0514, | |
| "step": 2455 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 4.052718286655684e-05, | |
| "loss": 0.0062, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "learning_rate": 4.060955518945634e-05, | |
| "loss": 0.1933, | |
| "step": 2465 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 4.069192751235585e-05, | |
| "loss": 0.0984, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 4.077429983525536e-05, | |
| "loss": 0.007, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 4.085667215815486e-05, | |
| "loss": 0.0773, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 4.093904448105437e-05, | |
| "loss": 0.163, | |
| "step": 2485 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "learning_rate": 4.102141680395387e-05, | |
| "loss": 0.0634, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 4.110378912685338e-05, | |
| "loss": 0.0718, | |
| "step": 2495 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "learning_rate": 4.118616144975289e-05, | |
| "loss": 0.0545, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 4.126853377265239e-05, | |
| "loss": 0.0109, | |
| "step": 2505 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 4.13509060955519e-05, | |
| "loss": 0.0373, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 4.14332784184514e-05, | |
| "loss": 0.0968, | |
| "step": 2515 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 4.1515650741350906e-05, | |
| "loss": 0.0431, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 4.159802306425041e-05, | |
| "loss": 0.0041, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 4.1680395387149915e-05, | |
| "loss": 0.0244, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.981, | |
| "eval_loss": 0.06960970908403397, | |
| "eval_runtime": 103.4462, | |
| "eval_samples_per_second": 29.001, | |
| "eval_steps_per_second": 1.817, | |
| "step": 2533 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 4.1762767710049426e-05, | |
| "loss": 0.1355, | |
| "step": 2535 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 4.184514003294893e-05, | |
| "loss": 0.0561, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 4.1927512355848435e-05, | |
| "loss": 0.0498, | |
| "step": 2545 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 4.200988467874794e-05, | |
| "loss": 0.0254, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 4.209225700164745e-05, | |
| "loss": 0.071, | |
| "step": 2555 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 4.2174629324546956e-05, | |
| "loss": 0.0298, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 4.225700164744646e-05, | |
| "loss": 0.0128, | |
| "step": 2565 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 4.2339373970345965e-05, | |
| "loss": 0.0994, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 4.2421746293245476e-05, | |
| "loss": 0.1421, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 9.17, | |
| "learning_rate": 4.250411861614498e-05, | |
| "loss": 0.1385, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 4.2586490939044485e-05, | |
| "loss": 0.0723, | |
| "step": 2585 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 4.266886326194399e-05, | |
| "loss": 0.0812, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 4.2751235584843494e-05, | |
| "loss": 0.0567, | |
| "step": 2595 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 4.2833607907743e-05, | |
| "loss": 0.0913, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "learning_rate": 4.29159802306425e-05, | |
| "loss": 0.062, | |
| "step": 2605 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "learning_rate": 4.299835255354201e-05, | |
| "loss": 0.0336, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "learning_rate": 4.308072487644152e-05, | |
| "loss": 0.0118, | |
| "step": 2615 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "learning_rate": 4.3163097199341024e-05, | |
| "loss": 0.0096, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 4.324546952224053e-05, | |
| "loss": 0.0836, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 4.332784184514003e-05, | |
| "loss": 0.0627, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 4.3410214168039544e-05, | |
| "loss": 0.0153, | |
| "step": 2635 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "learning_rate": 4.349258649093905e-05, | |
| "loss": 0.0043, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 4.357495881383855e-05, | |
| "loss": 0.0099, | |
| "step": 2645 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 4.365733113673806e-05, | |
| "loss": 0.0365, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 4.373970345963756e-05, | |
| "loss": 0.1346, | |
| "step": 2655 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 4.382207578253707e-05, | |
| "loss": 0.0603, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "learning_rate": 4.390444810543657e-05, | |
| "loss": 0.0286, | |
| "step": 2665 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 4.3986820428336076e-05, | |
| "loss": 0.004, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 4.406919275123559e-05, | |
| "loss": 0.0636, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 9.52, | |
| "learning_rate": 4.415156507413509e-05, | |
| "loss": 0.0452, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 4.4233937397034596e-05, | |
| "loss": 0.0294, | |
| "step": 2685 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 4.43163097199341e-05, | |
| "loss": 0.0908, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 4.439868204283361e-05, | |
| "loss": 0.0994, | |
| "step": 2695 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 4.448105436573312e-05, | |
| "loss": 0.0322, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 9.61, | |
| "learning_rate": 4.456342668863262e-05, | |
| "loss": 0.0027, | |
| "step": 2705 | |
| }, | |
| { | |
| "epoch": 9.63, | |
| "learning_rate": 4.464579901153213e-05, | |
| "loss": 0.0813, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 4.472817133443164e-05, | |
| "loss": 0.0263, | |
| "step": 2715 | |
| }, | |
| { | |
| "epoch": 9.66, | |
| "learning_rate": 4.481054365733114e-05, | |
| "loss": 0.0218, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 4.4892915980230646e-05, | |
| "loss": 0.0164, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "learning_rate": 4.497528830313015e-05, | |
| "loss": 0.0139, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 9.72, | |
| "learning_rate": 4.5057660626029655e-05, | |
| "loss": 0.0108, | |
| "step": 2735 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 4.514003294892916e-05, | |
| "loss": 0.0018, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 4.5222405271828664e-05, | |
| "loss": 0.0869, | |
| "step": 2745 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 4.5304777594728176e-05, | |
| "loss": 0.2074, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 9.79, | |
| "learning_rate": 4.538714991762768e-05, | |
| "loss": 0.1727, | |
| "step": 2755 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 4.5469522240527185e-05, | |
| "loss": 0.0816, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "learning_rate": 4.555189456342669e-05, | |
| "loss": 0.0424, | |
| "step": 2765 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 4.56342668863262e-05, | |
| "loss": 0.0198, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 4.5716639209225705e-05, | |
| "loss": 0.0544, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 4.579901153212521e-05, | |
| "loss": 0.0934, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 9.89, | |
| "learning_rate": 4.5881383855024714e-05, | |
| "loss": 0.0565, | |
| "step": 2785 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "learning_rate": 4.596375617792422e-05, | |
| "loss": 0.0963, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 4.6046128500823723e-05, | |
| "loss": 0.0417, | |
| "step": 2795 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 4.612850082372323e-05, | |
| "loss": 0.232, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 4.621087314662273e-05, | |
| "loss": 0.0678, | |
| "step": 2805 | |
| }, | |
| { | |
| "epoch": 9.98, | |
| "learning_rate": 4.6293245469522244e-05, | |
| "loss": 0.1453, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 4.637561779242175e-05, | |
| "loss": 0.0646, | |
| "step": 2815 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.995, | |
| "eval_loss": 0.011982438154518604, | |
| "eval_runtime": 103.5553, | |
| "eval_samples_per_second": 28.97, | |
| "eval_steps_per_second": 1.815, | |
| "step": 2815 | |
| }, | |
| { | |
| "epoch": 10.02, | |
| "learning_rate": 4.645799011532125e-05, | |
| "loss": 0.0381, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 4.654036243822076e-05, | |
| "loss": 0.0457, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "learning_rate": 4.662273476112027e-05, | |
| "loss": 0.0604, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 10.07, | |
| "learning_rate": 4.670510708401977e-05, | |
| "loss": 0.0741, | |
| "step": 2835 | |
| }, | |
| { | |
| "epoch": 10.09, | |
| "learning_rate": 4.678747940691928e-05, | |
| "loss": 0.011, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 10.11, | |
| "learning_rate": 4.686985172981878e-05, | |
| "loss": 0.1017, | |
| "step": 2845 | |
| }, | |
| { | |
| "epoch": 10.12, | |
| "learning_rate": 4.6952224052718294e-05, | |
| "loss": 0.0778, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 4.70345963756178e-05, | |
| "loss": 0.0241, | |
| "step": 2855 | |
| }, | |
| { | |
| "epoch": 10.16, | |
| "learning_rate": 4.71169686985173e-05, | |
| "loss": 0.0269, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 10.18, | |
| "learning_rate": 4.719934102141681e-05, | |
| "loss": 0.0835, | |
| "step": 2865 | |
| }, | |
| { | |
| "epoch": 10.2, | |
| "learning_rate": 4.728171334431631e-05, | |
| "loss": 0.0304, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 10.21, | |
| "learning_rate": 4.7364085667215816e-05, | |
| "loss": 0.0377, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 10.23, | |
| "learning_rate": 4.744645799011532e-05, | |
| "loss": 0.0199, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 10.25, | |
| "learning_rate": 4.7528830313014825e-05, | |
| "loss": 0.0047, | |
| "step": 2885 | |
| }, | |
| { | |
| "epoch": 10.27, | |
| "learning_rate": 4.761120263591434e-05, | |
| "loss": 0.0524, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 10.28, | |
| "learning_rate": 4.769357495881384e-05, | |
| "loss": 0.0325, | |
| "step": 2895 | |
| }, | |
| { | |
| "epoch": 10.3, | |
| "learning_rate": 4.7775947281713346e-05, | |
| "loss": 0.0856, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 10.32, | |
| "learning_rate": 4.785831960461285e-05, | |
| "loss": 0.0637, | |
| "step": 2905 | |
| }, | |
| { | |
| "epoch": 10.34, | |
| "learning_rate": 4.794069192751236e-05, | |
| "loss": 0.0531, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 10.36, | |
| "learning_rate": 4.8023064250411866e-05, | |
| "loss": 0.0931, | |
| "step": 2915 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "learning_rate": 4.810543657331137e-05, | |
| "loss": 0.0554, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "learning_rate": 4.8187808896210875e-05, | |
| "loss": 0.1008, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 10.41, | |
| "learning_rate": 4.827018121911038e-05, | |
| "loss": 0.0669, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 10.43, | |
| "learning_rate": 4.8352553542009884e-05, | |
| "loss": 0.0034, | |
| "step": 2935 | |
| }, | |
| { | |
| "epoch": 10.44, | |
| "learning_rate": 4.843492586490939e-05, | |
| "loss": 0.0409, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 10.46, | |
| "learning_rate": 4.8517298187808894e-05, | |
| "loss": 0.0513, | |
| "step": 2945 | |
| }, | |
| { | |
| "epoch": 10.48, | |
| "learning_rate": 4.8599670510708405e-05, | |
| "loss": 0.1203, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 10.5, | |
| "learning_rate": 4.868204283360791e-05, | |
| "loss": 0.0321, | |
| "step": 2955 | |
| }, | |
| { | |
| "epoch": 10.52, | |
| "learning_rate": 4.8764415156507414e-05, | |
| "loss": 0.077, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 4.884678747940692e-05, | |
| "loss": 0.0277, | |
| "step": 2965 | |
| }, | |
| { | |
| "epoch": 10.55, | |
| "learning_rate": 4.892915980230643e-05, | |
| "loss": 0.0299, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 10.57, | |
| "learning_rate": 4.9011532125205934e-05, | |
| "loss": 0.1544, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 10.59, | |
| "learning_rate": 4.909390444810544e-05, | |
| "loss": 0.0122, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 10.6, | |
| "learning_rate": 4.9176276771004943e-05, | |
| "loss": 0.1204, | |
| "step": 2985 | |
| }, | |
| { | |
| "epoch": 10.62, | |
| "learning_rate": 4.9258649093904455e-05, | |
| "loss": 0.0048, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 10.64, | |
| "learning_rate": 4.934102141680396e-05, | |
| "loss": 0.0579, | |
| "step": 2995 | |
| }, | |
| { | |
| "epoch": 10.66, | |
| "learning_rate": 4.9423393739703464e-05, | |
| "loss": 0.064, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 10.67, | |
| "learning_rate": 4.950576606260297e-05, | |
| "loss": 0.0405, | |
| "step": 3005 | |
| }, | |
| { | |
| "epoch": 10.69, | |
| "learning_rate": 4.958813838550247e-05, | |
| "loss": 0.0614, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 10.71, | |
| "learning_rate": 4.967051070840198e-05, | |
| "loss": 0.056, | |
| "step": 3015 | |
| }, | |
| { | |
| "epoch": 10.73, | |
| "learning_rate": 4.975288303130148e-05, | |
| "loss": 0.0657, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 10.75, | |
| "learning_rate": 4.9835255354200987e-05, | |
| "loss": 0.0233, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 10.76, | |
| "learning_rate": 4.99176276771005e-05, | |
| "loss": 0.0463, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 10.78, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0842, | |
| "step": 3035 | |
| }, | |
| { | |
| "epoch": 10.8, | |
| "learning_rate": 4.925816023738873e-05, | |
| "loss": 0.0374, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 10.82, | |
| "learning_rate": 4.851632047477745e-05, | |
| "loss": 0.0708, | |
| "step": 3045 | |
| }, | |
| { | |
| "epoch": 10.83, | |
| "learning_rate": 4.777448071216617e-05, | |
| "loss": 0.0392, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 4.70326409495549e-05, | |
| "loss": 0.1097, | |
| "step": 3055 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "learning_rate": 4.6290801186943624e-05, | |
| "loss": 0.0224, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 10.89, | |
| "learning_rate": 4.554896142433235e-05, | |
| "loss": 0.0549, | |
| "step": 3065 | |
| }, | |
| { | |
| "epoch": 10.91, | |
| "learning_rate": 4.480712166172107e-05, | |
| "loss": 0.0094, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 10.92, | |
| "learning_rate": 4.4065281899109794e-05, | |
| "loss": 0.0463, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 10.94, | |
| "learning_rate": 4.332344213649852e-05, | |
| "loss": 0.0399, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 10.96, | |
| "learning_rate": 4.258160237388724e-05, | |
| "loss": 0.0381, | |
| "step": 3085 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 4.1839762611275965e-05, | |
| "loss": 0.0571, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 10.99, | |
| "learning_rate": 4.109792284866469e-05, | |
| "loss": 0.0025, | |
| "step": 3095 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.9976666666666667, | |
| "eval_loss": 0.007643451914191246, | |
| "eval_runtime": 102.5928, | |
| "eval_samples_per_second": 29.242, | |
| "eval_steps_per_second": 1.832, | |
| "step": 3096 | |
| }, | |
| { | |
| "epoch": 11.01, | |
| "learning_rate": 4.0356083086053416e-05, | |
| "loss": 0.0152, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 3.961424332344214e-05, | |
| "loss": 0.0327, | |
| "step": 3105 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 3.887240356083086e-05, | |
| "loss": 0.026, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 11.07, | |
| "learning_rate": 3.8130563798219586e-05, | |
| "loss": 0.0184, | |
| "step": 3115 | |
| }, | |
| { | |
| "epoch": 11.08, | |
| "learning_rate": 3.7388724035608305e-05, | |
| "loss": 0.0115, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 11.1, | |
| "learning_rate": 3.664688427299703e-05, | |
| "loss": 0.0302, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 3.590504451038576e-05, | |
| "loss": 0.0561, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 11.14, | |
| "learning_rate": 3.516320474777448e-05, | |
| "loss": 0.0437, | |
| "step": 3135 | |
| }, | |
| { | |
| "epoch": 11.15, | |
| "learning_rate": 3.442136498516321e-05, | |
| "loss": 0.0563, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 11.17, | |
| "learning_rate": 3.3679525222551934e-05, | |
| "loss": 0.0032, | |
| "step": 3145 | |
| }, | |
| { | |
| "epoch": 11.19, | |
| "learning_rate": 3.293768545994065e-05, | |
| "loss": 0.0436, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 11.21, | |
| "learning_rate": 3.219584569732938e-05, | |
| "loss": 0.0518, | |
| "step": 3155 | |
| }, | |
| { | |
| "epoch": 11.23, | |
| "learning_rate": 3.14540059347181e-05, | |
| "loss": 0.0702, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 11.24, | |
| "learning_rate": 3.071216617210683e-05, | |
| "loss": 0.1229, | |
| "step": 3165 | |
| }, | |
| { | |
| "epoch": 11.26, | |
| "learning_rate": 2.9970326409495552e-05, | |
| "loss": 0.0024, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 11.28, | |
| "learning_rate": 2.9228486646884274e-05, | |
| "loss": 0.0739, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 11.3, | |
| "learning_rate": 2.8486646884273e-05, | |
| "loss": 0.0021, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 11.31, | |
| "learning_rate": 2.774480712166172e-05, | |
| "loss": 0.0408, | |
| "step": 3185 | |
| }, | |
| { | |
| "epoch": 11.33, | |
| "learning_rate": 2.7002967359050445e-05, | |
| "loss": 0.0087, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 11.35, | |
| "learning_rate": 2.6261127596439174e-05, | |
| "loss": 0.0829, | |
| "step": 3195 | |
| }, | |
| { | |
| "epoch": 11.37, | |
| "learning_rate": 2.5519287833827893e-05, | |
| "loss": 0.0153, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 11.39, | |
| "learning_rate": 2.4777448071216618e-05, | |
| "loss": 0.0131, | |
| "step": 3205 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 2.4035608308605344e-05, | |
| "loss": 0.0306, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 11.42, | |
| "learning_rate": 2.3293768545994066e-05, | |
| "loss": 0.0877, | |
| "step": 3215 | |
| }, | |
| { | |
| "epoch": 11.44, | |
| "learning_rate": 2.255192878338279e-05, | |
| "loss": 0.0418, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 11.46, | |
| "learning_rate": 2.1810089020771514e-05, | |
| "loss": 0.0028, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 11.47, | |
| "learning_rate": 2.106824925816024e-05, | |
| "loss": 0.0135, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "learning_rate": 2.0326409495548962e-05, | |
| "loss": 0.0013, | |
| "step": 3235 | |
| }, | |
| { | |
| "epoch": 11.51, | |
| "learning_rate": 1.9584569732937684e-05, | |
| "loss": 0.0112, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 11.53, | |
| "learning_rate": 1.884272997032641e-05, | |
| "loss": 0.0501, | |
| "step": 3245 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 1.8100890207715136e-05, | |
| "loss": 0.0022, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 11.56, | |
| "learning_rate": 1.7359050445103858e-05, | |
| "loss": 0.0012, | |
| "step": 3255 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "learning_rate": 1.661721068249258e-05, | |
| "loss": 0.0452, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 11.6, | |
| "learning_rate": 1.5875370919881306e-05, | |
| "loss": 0.0662, | |
| "step": 3265 | |
| }, | |
| { | |
| "epoch": 11.62, | |
| "learning_rate": 1.5133531157270032e-05, | |
| "loss": 0.0114, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 11.63, | |
| "learning_rate": 1.4391691394658754e-05, | |
| "loss": 0.0482, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 11.65, | |
| "learning_rate": 1.3649851632047478e-05, | |
| "loss": 0.0551, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 11.67, | |
| "learning_rate": 1.29080118694362e-05, | |
| "loss": 0.0155, | |
| "step": 3285 | |
| }, | |
| { | |
| "epoch": 11.69, | |
| "learning_rate": 1.2166172106824926e-05, | |
| "loss": 0.0455, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 11.71, | |
| "learning_rate": 1.142433234421365e-05, | |
| "loss": 0.0024, | |
| "step": 3295 | |
| }, | |
| { | |
| "epoch": 11.72, | |
| "learning_rate": 1.0682492581602374e-05, | |
| "loss": 0.0029, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 11.74, | |
| "learning_rate": 9.940652818991098e-06, | |
| "loss": 0.0191, | |
| "step": 3305 | |
| }, | |
| { | |
| "epoch": 11.76, | |
| "learning_rate": 9.198813056379822e-06, | |
| "loss": 0.0294, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 11.78, | |
| "learning_rate": 8.456973293768548e-06, | |
| "loss": 0.0277, | |
| "step": 3315 | |
| }, | |
| { | |
| "epoch": 11.79, | |
| "learning_rate": 7.71513353115727e-06, | |
| "loss": 0.0462, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 6.973293768545995e-06, | |
| "loss": 0.0234, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 11.83, | |
| "learning_rate": 6.231454005934718e-06, | |
| "loss": 0.0012, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 11.85, | |
| "learning_rate": 5.489614243323442e-06, | |
| "loss": 0.0045, | |
| "step": 3335 | |
| }, | |
| { | |
| "epoch": 11.87, | |
| "learning_rate": 4.747774480712167e-06, | |
| "loss": 0.0314, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "learning_rate": 4.005934718100891e-06, | |
| "loss": 0.0378, | |
| "step": 3345 | |
| }, | |
| { | |
| "epoch": 11.9, | |
| "learning_rate": 3.2640949554896145e-06, | |
| "loss": 0.0012, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 11.92, | |
| "learning_rate": 2.5222551928783385e-06, | |
| "loss": 0.0756, | |
| "step": 3355 | |
| }, | |
| { | |
| "epoch": 11.94, | |
| "learning_rate": 1.7804154302670625e-06, | |
| "loss": 0.0021, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 11.95, | |
| "learning_rate": 1.0385756676557863e-06, | |
| "loss": 0.0044, | |
| "step": 3365 | |
| }, | |
| { | |
| "epoch": 11.97, | |
| "learning_rate": 2.9673590504451043e-07, | |
| "loss": 0.0611, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 11.98, | |
| "eval_accuracy": 0.9996666666666667, | |
| "eval_loss": 0.0023621288128197193, | |
| "eval_runtime": 102.266, | |
| "eval_samples_per_second": 29.335, | |
| "eval_steps_per_second": 1.838, | |
| "step": 3372 | |
| }, | |
| { | |
| "epoch": 11.98, | |
| "step": 3372, | |
| "total_flos": 1.9127019697307763e+19, | |
| "train_loss": 0.1604562966202188, | |
| "train_runtime": 12187.6968, | |
| "train_samples_per_second": 8.861, | |
| "train_steps_per_second": 0.277 | |
| } | |
| ], | |
| "max_steps": 3372, | |
| "num_train_epochs": 12, | |
| "total_flos": 1.9127019697307763e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |