| { | |
| "best_metric": 1.4596115350723267, | |
| "best_model_checkpoint": "output/boris-grebenshikov/checkpoint-1903", | |
| "epoch": 11.0, | |
| "global_step": 1903, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00013690389360668606, | |
| "loss": 2.5933, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00013601813066569938, | |
| "loss": 2.3572, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00013455035782630487, | |
| "loss": 2.2816, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00013251324613578177, | |
| "loss": 2.2503, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00012992438165230672, | |
| "loss": 2.0864, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00012680611362733682, | |
| "loss": 2.1159, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00012318536156811003, | |
| "loss": 2.0993, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00011909338284586077, | |
| "loss": 1.8802, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00011456550285595239, | |
| "loss": 2.0531, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00010964081005941026, | |
| "loss": 1.9281, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0001043618185385132, | |
| "loss": 2.1099, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.877410097954656e-05, | |
| "loss": 2.0119, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.292589525111794e-05, | |
| "loss": 1.9675, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.686768797438203e-05, | |
| "loss": 1.9512, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.065177868014528e-05, | |
| "loss": 2.047, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.433182831541305e-05, | |
| "loss": 1.9436, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 6.796239599704895e-05, | |
| "loss": 1.9677, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.159846801167835e-05, | |
| "loss": 1.8582, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 5.529498312790352e-05, | |
| "loss": 1.9952, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.9106358318734425e-05, | |
| "loss": 1.9317, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.3086018988597235e-05, | |
| "loss": 1.834, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.728593776039493e-05, | |
| "loss": 1.8923, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.1756185804197785e-05, | |
| "loss": 1.9544, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.6544500580870892e-05, | |
| "loss": 1.921, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.169587373223826e-05, | |
| "loss": 1.9539, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.725216267546246e-05, | |
| "loss": 1.9665, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.3251729254682012e-05, | |
| "loss": 2.0147, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.729108569369736e-06, | |
| "loss": 1.8575, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.7147108383636075e-06, | |
| "loss": 1.8372, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.234558873329575e-06, | |
| "loss": 1.7437, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.310063428006295e-06, | |
| "loss": 1.8886, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.578383626055595e-07, | |
| "loss": 1.9129, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.895572190242788e-07, | |
| "loss": 1.8767, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.8803279399871826, | |
| "eval_runtime": 11.5804, | |
| "eval_samples_per_second": 22.279, | |
| "eval_steps_per_second": 2.85, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0177397892100256e-07, | |
| "loss": 1.9274, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.744057536125183, | |
| "eval_runtime": 10.1523, | |
| "eval_samples_per_second": 22.458, | |
| "eval_steps_per_second": 2.856, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.523909530405488e-08, | |
| "loss": 1.937, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 5.534939285797931e-07, | |
| "loss": 1.8799, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.622351211524088e-06, | |
| "loss": 1.8935, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.243005142843674e-06, | |
| "loss": 1.873, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.402103933939619e-06, | |
| "loss": 1.8993, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.081859807874568e-06, | |
| "loss": 1.9433, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.1260195544315306e-05, | |
| "loss": 1.8802, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.491092636313743e-05, | |
| "loss": 1.9911, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.90039756482401e-05, | |
| "loss": 1.9398, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.350562273432663e-05, | |
| "loss": 1.7847, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8378780715254437e-05, | |
| "loss": 1.8382, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.358330198522479e-05, | |
| "loss": 1.9425, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.907630899560607e-05, | |
| "loss": 1.9129, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.4812547502442855e-05, | |
| "loss": 1.8725, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 5.0744759394415807e-05, | |
| "loss": 1.9258, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 5.682407202970108e-05, | |
| "loss": 1.8416, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 6.300040087417042e-05, | |
| "loss": 1.8097, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 6.922286212378929e-05, | |
| "loss": 1.8724, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.544019191181583e-05, | |
| "loss": 1.8985, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.160116864715307e-05, | |
| "loss": 1.877, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.765503500441301e-05, | |
| "loss": 1.8627, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.35519160891186e-05, | |
| "loss": 1.7856, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 9.92432303329815e-05, | |
| "loss": 1.8525, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00010468208973408737, | |
| "loss": 1.9036, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00010982368614460176, | |
| "loss": 1.8999, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00011462566042355846, | |
| "loss": 1.866, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00011904845141345724, | |
| "loss": 1.9433, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00012305562186562137, | |
| "loss": 1.8415, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0001266141586291718, | |
| "loss": 1.8489, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.00012969474463050626, | |
| "loss": 1.7954, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00013227200040257384, | |
| "loss": 1.8456, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.000134324693174103, | |
| "loss": 1.8414, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00013583591179619617, | |
| "loss": 1.8602, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.00013679320606515522, | |
| "loss": 1.8914, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.00013718868929372445, | |
| "loss": 1.7701, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.7017050981521606, | |
| "eval_runtime": 10.2794, | |
| "eval_samples_per_second": 22.18, | |
| "eval_steps_per_second": 2.821, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.00013603195463831566, | |
| "loss": 1.8257, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.00013458123912165538, | |
| "loss": 1.7795, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.00013256759493713883, | |
| "loss": 1.6858, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.0001300082017869573, | |
| "loss": 1.8715, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.00012692489551105156, | |
| "loss": 1.7653, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.0001233439817914244, | |
| "loss": 1.7267, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.0001192960117213372, | |
| "loss": 1.7223, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.00011481552115415387, | |
| "loss": 1.7739, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 0.00010994073605561706, | |
| "loss": 1.7485, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.00010471324637338657, | |
| "loss": 1.7833, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 9.917765120627052e-05, | |
| "loss": 1.7504, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 9.338117830043871e-05, | |
| "loss": 1.7411, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 8.737328111894491e-05, | |
| "loss": 1.7067, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 8.120521692221671e-05, | |
| "loss": 1.7285, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.492960945918252e-05, | |
| "loss": 1.7692, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.860000000000001e-05, | |
| "loss": 1.7354, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.227039054081752e-05, | |
| "loss": 1.7446, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 5.599478307778333e-05, | |
| "loss": 1.8284, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.9826718881055135e-05, | |
| "loss": 1.7434, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.381882169956128e-05, | |
| "loss": 1.6965, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 3.8022348793729525e-05, | |
| "loss": 1.7405, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 3.24867536266134e-05, | |
| "loss": 1.7292, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 2.7259263944382986e-05, | |
| "loss": 1.7385, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 2.2384478845846205e-05, | |
| "loss": 1.6668, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.7903988278662788e-05, | |
| "loss": 1.591, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.3856018208575617e-05, | |
| "loss": 1.7083, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.0275104488948488e-05, | |
| "loss": 1.6955, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 7.191798213042723e-06, | |
| "loss": 1.7392, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.6324050628612214e-06, | |
| "loss": 1.7917, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 2.6187608783446213e-06, | |
| "loss": 1.7505, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.1680453616843376e-06, | |
| "loss": 1.7224, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.926355061606279e-07, | |
| "loss": 1.6373, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.6567, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 1.6859837770462036, | |
| "eval_runtime": 10.9838, | |
| "eval_samples_per_second": 22.761, | |
| "eval_steps_per_second": 2.913, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.8089671428491253e-07, | |
| "loss": 1.7435, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 1.5615885257720947, | |
| "eval_runtime": 14.4733, | |
| "eval_samples_per_second": 15.753, | |
| "eval_steps_per_second": 2.004, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 1.1310706275549354e-08, | |
| "loss": 1.7668, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.0679393484475955e-07, | |
| "loss": 1.6029, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.3640882038038144e-06, | |
| "loss": 1.7397, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.875306825896957e-06, | |
| "loss": 1.6337, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.927999597426121e-06, | |
| "loss": 1.7185, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 7.505255369493709e-06, | |
| "loss": 1.7042, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.0585841370828143e-05, | |
| "loss": 1.6665, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.4144378134378619e-05, | |
| "loss": 1.6517, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.8151548586542735e-05, | |
| "loss": 1.7248, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 2.257433957644151e-05, | |
| "loss": 1.7293, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 2.7376313855398193e-05, | |
| "loss": 1.5925, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.25179102659126e-05, | |
| "loss": 1.7049, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.795676966701845e-05, | |
| "loss": 1.6687, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 4.3648083910881356e-05, | |
| "loss": 1.6983, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 4.954496499558694e-05, | |
| "loss": 1.785, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 5.559883135284689e-05, | |
| "loss": 1.698, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 6.175980808818411e-05, | |
| "loss": 1.6591, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 6.797713787621067e-05, | |
| "loss": 1.7702, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.419959912582961e-05, | |
| "loss": 1.7052, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 8.037592797029894e-05, | |
| "loss": 1.6658, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 8.645524060558421e-05, | |
| "loss": 1.6903, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 9.238745249755704e-05, | |
| "loss": 1.7102, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 9.812369100439384e-05, | |
| "loss": 1.7289, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 0.00010361669801477512, | |
| "loss": 1.688, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 0.00010882121928474553, | |
| "loss": 1.7392, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 0.00011369437726567332, | |
| "loss": 1.7228, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 0.00011819602435175987, | |
| "loss": 1.7298, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 0.0001222890736368625, | |
| "loss": 1.6633, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 0.00012593980445568464, | |
| "loss": 1.7014, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 0.0001291181401921254, | |
| "loss": 1.6952, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 0.00013179789606606037, | |
| "loss": 1.7324, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 0.0001339569948571563, | |
| "loss": 1.6753, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 0.00013557764878847588, | |
| "loss": 1.834, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 0.0001366465060714202, | |
| "loss": 1.6832, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 0.00013715476090469595, | |
| "loss": 1.703, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 1.588655710220337, | |
| "eval_runtime": 14.8949, | |
| "eval_samples_per_second": 15.307, | |
| "eval_steps_per_second": 1.947, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 0.000137098226021079, | |
| "loss": 1.6012, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 0.0001364773671842959, | |
| "loss": 1.7298, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 0.0001352972993518206, | |
| "loss": 1.6923, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 0.0001335677445351985, | |
| "loss": 1.7106, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 0.00013130295170506775, | |
| "loss": 1.6598, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 0.00012852157940074013, | |
| "loss": 1.6285, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 0.00012524654201146593, | |
| "loss": 1.6431, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 0.00012150482099579733, | |
| "loss": 1.5908, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 0.00011732724259432273, | |
| "loss": 1.6991, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 0.00011274822386709064, | |
| "loss": 1.5654, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 0.00010780548914799808, | |
| "loss": 1.6367, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 0.00010253975925213988, | |
| "loss": 1.7132, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 9.699441599659e-05, | |
| "loss": 1.6216, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 9.121514479846673e-05, | |
| "loss": 1.6058, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 8.52495582947446e-05, | |
| "loss": 1.5635, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 7.914680408462862e-05, | |
| "loss": 1.6392, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 7.295715982611233e-05, | |
| "loss": 1.5928, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 6.673161902252507e-05, | |
| "loss": 1.5735, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 6.052147091157344e-05, | |
| "loss": 1.645, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 5.4377877917969595e-05, | |
| "loss": 1.6497, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 4.8351454150810734e-05, | |
| "loss": 1.6139, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 4.249184841826729e-05, | |
| "loss": 1.5811, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.6847335194921544e-05, | |
| "loss": 1.5699, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.146441691157973e-05, | |
| "loss": 1.6003, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 2.638744084410121e-05, | |
| "loss": 1.6607, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.165823375751248e-05, | |
| "loss": 1.6131, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 1.7315757315399247e-05, | |
| "loss": 1.5716, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 1.3395787093490896e-05, | |
| "loss": 1.6196, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 9.930617841891443e-06, | |
| "loss": 1.6354, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 6.948797424159463e-06, | |
| "loss": 1.5669, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 4.474891625184719e-06, | |
| "loss": 1.604, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 2.5292817654967284e-06, | |
| "loss": 1.523, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 1.127996789363344e-06, | |
| "loss": 1.6719, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 2.8258121002494224e-07, | |
| "loss": 1.6116, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.0, | |
| "loss": 1.4979, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 1.516666054725647, | |
| "eval_runtime": 14.812, | |
| "eval_samples_per_second": 15.393, | |
| "eval_steps_per_second": 1.958, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.825812100249346e-07, | |
| "loss": 1.5673, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 1.127996789363329e-06, | |
| "loss": 1.5619, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.529281765496713e-06, | |
| "loss": 1.5238, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 4.4748916251846885e-06, | |
| "loss": 1.5144, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 6.948797424159478e-06, | |
| "loss": 1.5482, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 9.930617841891406e-06, | |
| "loss": 1.593, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 1.3395787093490912e-05, | |
| "loss": 1.6012, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 1.7315757315399193e-05, | |
| "loss": 1.5677, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 2.1658233757512504e-05, | |
| "loss": 1.5793, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 2.638744084410115e-05, | |
| "loss": 1.5031, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 3.1464416911579663e-05, | |
| "loss": 1.5173, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 3.684733519492136e-05, | |
| "loss": 1.489, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 4.249184841826722e-05, | |
| "loss": 1.6251, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 4.835145415081054e-05, | |
| "loss": 1.5882, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 5.4377877917969514e-05, | |
| "loss": 1.6055, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 6.052147091157324e-05, | |
| "loss": 1.5492, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 6.673161902252498e-05, | |
| "loss": 1.534, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 7.295715982611214e-05, | |
| "loss": 1.5354, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 7.914680408462853e-05, | |
| "loss": 1.6055, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 8.52495582947444e-05, | |
| "loss": 1.5166, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 9.121514479846663e-05, | |
| "loss": 1.5157, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 9.699441599658982e-05, | |
| "loss": 1.6171, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 0.00010253975925213981, | |
| "loss": 1.5676, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 0.0001078054891479979, | |
| "loss": 1.5382, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 0.00011274822386709058, | |
| "loss": 1.6057, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 0.00011732724259432258, | |
| "loss": 1.5982, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 0.00012150482099579729, | |
| "loss": 1.5857, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 0.00012524654201146596, | |
| "loss": 1.6016, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 0.00012852157940074007, | |
| "loss": 1.552, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 0.00013130295170506778, | |
| "loss": 1.6309, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 0.00013356774453519848, | |
| "loss": 1.6005, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 0.0001352972993518206, | |
| "loss": 1.5484, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 0.00013647736718429588, | |
| "loss": 1.5604, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 0.000137098226021079, | |
| "loss": 1.5962, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 1.5382394790649414, | |
| "eval_runtime": 14.8295, | |
| "eval_samples_per_second": 15.375, | |
| "eval_steps_per_second": 1.956, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 0.00013715476090469595, | |
| "loss": 1.5942, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 0.0001366465060714202, | |
| "loss": 1.4645, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 0.00013557764878847597, | |
| "loss": 1.5359, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 0.00013395699485715633, | |
| "loss": 1.5554, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 0.00013179789606606045, | |
| "loss": 1.5214, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 0.00012911814019212547, | |
| "loss": 1.4883, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 0.00012593980445568475, | |
| "loss": 1.536, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 6.21, | |
| "learning_rate": 0.00012228907363686257, | |
| "loss": 1.4736, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 0.00011819602435175999, | |
| "loss": 1.5588, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 0.00011369437726567339, | |
| "loss": 1.6177, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 0.00010882121928474568, | |
| "loss": 1.5433, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 0.00010361669801477519, | |
| "loss": 1.4821, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 9.812369100439403e-05, | |
| "loss": 1.5071, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 9.238745249755713e-05, | |
| "loss": 1.5067, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 8.645524060558428e-05, | |
| "loss": 1.5131, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 8.037592797029891e-05, | |
| "loss": 1.5808, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 7.419959912582969e-05, | |
| "loss": 1.5181, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 6.797713787621063e-05, | |
| "loss": 1.5158, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 6.17598080881842e-05, | |
| "loss": 1.5602, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "learning_rate": 5.559883135284685e-05, | |
| "loss": 1.5248, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 4.954496499558702e-05, | |
| "loss": 1.6044, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 4.364808391088155e-05, | |
| "loss": 1.5675, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "learning_rate": 3.795676966701853e-05, | |
| "loss": 1.5317, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 3.2517910265912777e-05, | |
| "loss": 1.498, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 2.737631385539826e-05, | |
| "loss": 1.5435, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "learning_rate": 2.2574339576441664e-05, | |
| "loss": 1.4874, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 1.815154858654279e-05, | |
| "loss": 1.4665, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 1.4144378134378748e-05, | |
| "loss": 1.522, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 1.0585841370828188e-05, | |
| "loss": 1.4874, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 7.505255369493801e-06, | |
| "loss": 1.4275, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 6.88, | |
| "learning_rate": 4.927999597426152e-06, | |
| "loss": 1.5011, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 2.8753068258970103e-06, | |
| "loss": 1.5038, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 6.94, | |
| "learning_rate": 1.3640882038038297e-06, | |
| "loss": 1.5051, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 4.067939348447824e-07, | |
| "loss": 1.508, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "learning_rate": 1.1310706275549354e-08, | |
| "loss": 1.5261, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_loss": 1.4838725328445435, | |
| "eval_runtime": 14.6674, | |
| "eval_samples_per_second": 15.545, | |
| "eval_steps_per_second": 1.977, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 1.808967142848973e-07, | |
| "loss": 1.4801, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 9.141548212825557e-07, | |
| "loss": 1.4927, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 2.2050440662441005e-06, | |
| "loss": 1.4083, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 4.0429294332848074e-06, | |
| "loss": 1.3969, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 6.412669468442276e-06, | |
| "loss": 1.3985, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 9.294741022836825e-06, | |
| "loss": 1.4522, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 1.266540009450849e-05, | |
| "loss": 1.4855, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 1.649687744383478e-05, | |
| "loss": 1.3859, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 2.075760737094545e-05, | |
| "loss": 1.5523, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 2.5412487770347686e-05, | |
| "loss": 1.4851, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 3.0423169320298854e-05, | |
| "loss": 1.4955, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 3.57483714244302e-05, | |
| "loss": 1.4801, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 4.134422230273594e-05, | |
| "loss": 1.3948, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 4.7164620430075375e-05, | |
| "loss": 1.5297, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 5.316161434447666e-05, | |
| "loss": 1.5011, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 5.92857976961771e-05, | |
| "loss": 1.4813, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 6.548671628278832e-05, | |
| "loss": 1.476, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "learning_rate": 7.171328371721156e-05, | |
| "loss": 1.3971, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 7.791420230382278e-05, | |
| "loss": 1.4009, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 8.403838565552322e-05, | |
| "loss": 1.4851, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 9.00353795699245e-05, | |
| "loss": 1.4137, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "learning_rate": 9.585577769726395e-05, | |
| "loss": 1.4647, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 0.00010145162857556969, | |
| "loss": 1.3558, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 0.00010677683067970103, | |
| "loss": 1.4209, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "learning_rate": 0.00011178751222965221, | |
| "loss": 1.5359, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 0.00011644239262905447, | |
| "loss": 1.5243, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 0.00012070312255616514, | |
| "loss": 1.5307, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 0.00012453459990549145, | |
| "loss": 1.3905, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 0.0001279052589771631, | |
| "loss": 1.4955, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 0.00013078733053155768, | |
| "loss": 1.501, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 0.00013315707056671514, | |
| "loss": 1.4948, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 0.00013499495593375586, | |
| "loss": 1.496, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 0.00013628584517871742, | |
| "loss": 1.4929, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 7.98, | |
| "learning_rate": 0.0001370191032857151, | |
| "loss": 1.5077, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 1.5043175220489502, | |
| "eval_runtime": 14.7576, | |
| "eval_samples_per_second": 15.45, | |
| "eval_steps_per_second": 1.965, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 0.00013718868929372445, | |
| "loss": 1.5034, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 0.00013679320606515522, | |
| "loss": 1.4037, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 0.00013583591179619614, | |
| "loss": 1.4957, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 8.09, | |
| "learning_rate": 0.000134324693174103, | |
| "loss": 1.5076, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 0.0001322720004025738, | |
| "loss": 1.503, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 0.00012969474463050637, | |
| "loss": 1.4043, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 0.00012661415862917202, | |
| "loss": 1.4017, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 0.00012305562186562148, | |
| "loss": 1.4622, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 0.00011904845141345745, | |
| "loss": 1.4555, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 0.00011462566042355861, | |
| "loss": 1.474, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 0.00010982368614460204, | |
| "loss": 1.4798, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 8.32, | |
| "learning_rate": 0.00010468208973408754, | |
| "loss": 1.4409, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 8.35, | |
| "learning_rate": 9.924323033298181e-05, | |
| "loss": 1.4283, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 9.35519160891188e-05, | |
| "loss": 1.4248, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 8.76550350044131e-05, | |
| "loss": 1.4627, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 8.160116864715327e-05, | |
| "loss": 1.5012, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 7.544019191181593e-05, | |
| "loss": 1.3866, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 6.92228621237895e-05, | |
| "loss": 1.4909, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 8.53, | |
| "learning_rate": 6.300040087417043e-05, | |
| "loss": 1.385, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 5.682407202970122e-05, | |
| "loss": 1.4766, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 5.0744759394415834e-05, | |
| "loss": 1.4607, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 4.481254750244299e-05, | |
| "loss": 1.4235, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 3.907630899560609e-05, | |
| "loss": 1.4085, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 3.358330198522491e-05, | |
| "loss": 1.4548, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 2.8378780715254406e-05, | |
| "loss": 1.4757, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 2.3505622734326714e-05, | |
| "loss": 1.3891, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "learning_rate": 1.900397564824009e-05, | |
| "loss": 1.3819, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 1.4910926363137506e-05, | |
| "loss": 1.4202, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 8.82, | |
| "learning_rate": 1.1260195544315306e-05, | |
| "loss": 1.3403, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 8.081859807874605e-06, | |
| "loss": 1.4776, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 5.402103933939604e-06, | |
| "loss": 1.4531, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 3.2430051428437045e-06, | |
| "loss": 1.373, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 1.6223512115240727e-06, | |
| "loss": 1.3284, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 5.534939285798083e-07, | |
| "loss": 1.3615, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 4.523909530404726e-08, | |
| "loss": 1.3743, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_loss": 1.464684247970581, | |
| "eval_runtime": 15.3121, | |
| "eval_samples_per_second": 14.89, | |
| "eval_steps_per_second": 1.894, | |
| "step": 1557 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 1.0177397892099493e-07, | |
| "loss": 1.4264, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 7.226328157040877e-07, | |
| "loss": 1.3427, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 9.08, | |
| "learning_rate": 1.9027006481794198e-06, | |
| "loss": 1.4135, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 9.1, | |
| "learning_rate": 3.6322554648015487e-06, | |
| "loss": 1.3408, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 5.897048294932268e-06, | |
| "loss": 1.355, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "learning_rate": 8.678420599259977e-06, | |
| "loss": 1.3903, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 1.1953457988534113e-05, | |
| "loss": 1.3435, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 1.5695179004202806e-05, | |
| "loss": 1.4126, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "learning_rate": 1.9872757405677322e-05, | |
| "loss": 1.3711, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 2.445177613290951e-05, | |
| "loss": 1.4319, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "learning_rate": 2.9394510852001993e-05, | |
| "loss": 1.3731, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 3.46602407478603e-05, | |
| "loss": 1.3164, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 4.020558400341007e-05, | |
| "loss": 1.3927, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "learning_rate": 4.598485520153301e-05, | |
| "loss": 1.3304, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 5.195044170525524e-05, | |
| "loss": 1.3834, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 5.80531959153711e-05, | |
| "loss": 1.4126, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 6.424284017388749e-05, | |
| "loss": 1.3473, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 7.046838097747464e-05, | |
| "loss": 1.299, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 7.66785290884264e-05, | |
| "loss": 1.3673, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 8.282212208203013e-05, | |
| "loss": 1.3913, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 8.884854584918912e-05, | |
| "loss": 1.3892, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 9.470815158173245e-05, | |
| "loss": 1.3557, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 0.00010035266480507833, | |
| "loss": 1.396, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 0.00010573558308842004, | |
| "loss": 1.3608, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "learning_rate": 0.00011081255915589876, | |
| "loss": 1.3399, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 0.00011554176624248741, | |
| "loss": 1.3982, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 0.00011988424268460073, | |
| "loss": 1.4392, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 0.000123804212906509, | |
| "loss": 1.4083, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 9.83, | |
| "learning_rate": 0.00012726938215810852, | |
| "loss": 1.434, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 0.00013025120257584047, | |
| "loss": 1.3252, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 0.00013272510837481527, | |
| "loss": 1.3899, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "learning_rate": 0.00013467071823450324, | |
| "loss": 1.3883, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 9.94, | |
| "learning_rate": 0.00013607200321063664, | |
| "loss": 1.4395, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 0.00013691741878997505, | |
| "loss": 1.4208, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 0.0001372, | |
| "loss": 1.4499, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_loss": 1.4852185249328613, | |
| "eval_runtime": 15.5041, | |
| "eval_samples_per_second": 14.706, | |
| "eval_steps_per_second": 1.87, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 0.00013691741878997505, | |
| "loss": 1.3342, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 0.0001360720032106367, | |
| "loss": 1.3537, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 10.09, | |
| "learning_rate": 0.00013467071823450327, | |
| "loss": 1.376, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 10.12, | |
| "learning_rate": 0.00013272510837481532, | |
| "loss": 1.3746, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 0.00013025120257584055, | |
| "loss": 1.3632, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 10.17, | |
| "learning_rate": 0.00012726938215810863, | |
| "loss": 1.4038, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 10.2, | |
| "learning_rate": 0.00012380421290650925, | |
| "loss": 1.3788, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 10.23, | |
| "learning_rate": 0.00011988424268460084, | |
| "loss": 1.3642, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 10.26, | |
| "learning_rate": 0.00011554176624248753, | |
| "loss": 1.3539, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 10.29, | |
| "learning_rate": 0.00011081255915589869, | |
| "loss": 1.371, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 10.32, | |
| "learning_rate": 0.00010573558308842038, | |
| "loss": 1.3995, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 10.35, | |
| "learning_rate": 0.00010035266480507848, | |
| "loss": 1.4698, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 10.38, | |
| "learning_rate": 9.47081515817326e-05, | |
| "loss": 1.3803, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 10.4, | |
| "learning_rate": 8.884854584918904e-05, | |
| "loss": 1.4079, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 10.43, | |
| "learning_rate": 8.282212208203053e-05, | |
| "loss": 1.4, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 10.46, | |
| "learning_rate": 7.667852908842657e-05, | |
| "loss": 1.3182, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 10.49, | |
| "learning_rate": 7.04683809774748e-05, | |
| "loss": 1.3298, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 10.52, | |
| "learning_rate": 6.424284017388742e-05, | |
| "loss": 1.348, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 10.55, | |
| "learning_rate": 5.8053195915371506e-05, | |
| "loss": 1.3953, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 10.58, | |
| "learning_rate": 5.1950441705255396e-05, | |
| "loss": 1.3451, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 10.61, | |
| "learning_rate": 4.5984855201533164e-05, | |
| "loss": 1.3076, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 10.64, | |
| "learning_rate": 4.0205584003410225e-05, | |
| "loss": 1.4287, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 10.66, | |
| "learning_rate": 3.4660240747860236e-05, | |
| "loss": 1.3216, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 10.69, | |
| "learning_rate": 2.939451085200233e-05, | |
| "loss": 1.341, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "learning_rate": 2.4451776132909644e-05, | |
| "loss": 1.3848, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 10.75, | |
| "learning_rate": 1.9872757405677438e-05, | |
| "loss": 1.328, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 10.78, | |
| "learning_rate": 1.5695179004202758e-05, | |
| "loss": 1.3147, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 1.1953457988534349e-05, | |
| "loss": 1.2886, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 10.84, | |
| "learning_rate": 8.678420599260062e-06, | |
| "loss": 1.2943, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "learning_rate": 5.8970482949323365e-06, | |
| "loss": 1.3852, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 10.9, | |
| "learning_rate": 3.6322554648015263e-06, | |
| "loss": 1.3121, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 10.92, | |
| "learning_rate": 1.902700648179519e-06, | |
| "loss": 1.37, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 10.95, | |
| "learning_rate": 7.226328157041105e-07, | |
| "loss": 1.3568, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 1.0177397892100256e-07, | |
| "loss": 1.2291, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_loss": 1.4596115350723267, | |
| "eval_runtime": 15.3008, | |
| "eval_samples_per_second": 14.901, | |
| "eval_steps_per_second": 1.895, | |
| "step": 1903 | |
| } | |
| ], | |
| "max_steps": 2249, | |
| "num_train_epochs": 13, | |
| "total_flos": 1979287142400000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |