| { | |
| "best_global_step": 1000, | |
| "best_metric": 1.0913933515548706, | |
| "best_model_checkpoint": "output/reasoning-model_v11/checkpoint-1000", | |
| "epoch": 0.37512895057676077, | |
| "eval_steps": 500, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0018756447528838038, | |
| "grad_norm": 17.375, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.5042, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0037512895057676076, | |
| "grad_norm": 14.875, | |
| "learning_rate": 2.25e-06, | |
| "loss": 1.413, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.005626934258651411, | |
| "grad_norm": 13.375, | |
| "learning_rate": 3.5e-06, | |
| "loss": 1.5803, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.007502579011535215, | |
| "grad_norm": 9.1875, | |
| "learning_rate": 4.75e-06, | |
| "loss": 1.5081, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.00937822376441902, | |
| "grad_norm": 9.5, | |
| "learning_rate": 6e-06, | |
| "loss": 1.3924, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.011253868517302822, | |
| "grad_norm": 7.96875, | |
| "learning_rate": 7.25e-06, | |
| "loss": 1.4063, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.013129513270186627, | |
| "grad_norm": 7.78125, | |
| "learning_rate": 8.5e-06, | |
| "loss": 1.2718, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01500515802307043, | |
| "grad_norm": 10.4375, | |
| "learning_rate": 9.75e-06, | |
| "loss": 1.2445, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.016880802775954235, | |
| "grad_norm": 7.59375, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 1.1666, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.01875644752883804, | |
| "grad_norm": 7.0, | |
| "learning_rate": 1.2250000000000001e-05, | |
| "loss": 1.2257, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02063209228172184, | |
| "grad_norm": 8.5625, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "loss": 1.1071, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.022507737034605645, | |
| "grad_norm": 7.5625, | |
| "learning_rate": 1.4750000000000003e-05, | |
| "loss": 1.2369, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02438338178748945, | |
| "grad_norm": 8.5625, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 1.2355, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.026259026540373254, | |
| "grad_norm": 9.8125, | |
| "learning_rate": 1.7250000000000003e-05, | |
| "loss": 1.0905, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.028134671293257058, | |
| "grad_norm": 6.6875, | |
| "learning_rate": 1.8500000000000002e-05, | |
| "loss": 1.1362, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03001031604614086, | |
| "grad_norm": 12.875, | |
| "learning_rate": 1.9750000000000002e-05, | |
| "loss": 1.1355, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.031885960799024664, | |
| "grad_norm": 8.0625, | |
| "learning_rate": 1.999988193210057e-05, | |
| "loss": 1.2262, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03376160555190847, | |
| "grad_norm": 6.40625, | |
| "learning_rate": 1.9999402286037404e-05, | |
| "loss": 1.0978, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03563725030479227, | |
| "grad_norm": 9.375, | |
| "learning_rate": 1.9998553700250286e-05, | |
| "loss": 1.098, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.03751289505767608, | |
| "grad_norm": 7.21875, | |
| "learning_rate": 1.9997336206048778e-05, | |
| "loss": 1.2503, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03938853981055988, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 1.999574984835377e-05, | |
| "loss": 1.0663, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.04126418456344368, | |
| "grad_norm": 7.46875, | |
| "learning_rate": 1.9993794685695792e-05, | |
| "loss": 1.3083, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04313982931632749, | |
| "grad_norm": 6.9375, | |
| "learning_rate": 1.9991470790212877e-05, | |
| "loss": 1.1906, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.04501547406921129, | |
| "grad_norm": 9.8125, | |
| "learning_rate": 1.9988778247647887e-05, | |
| "loss": 1.1358, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.046891118822095096, | |
| "grad_norm": 9.375, | |
| "learning_rate": 1.9985717157345346e-05, | |
| "loss": 1.1421, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.0487667635749789, | |
| "grad_norm": 7.125, | |
| "learning_rate": 1.998228763224779e-05, | |
| "loss": 1.1236, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0506424083278627, | |
| "grad_norm": 6.1875, | |
| "learning_rate": 1.9978489798891584e-05, | |
| "loss": 1.2223, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.05251805308074651, | |
| "grad_norm": 6.875, | |
| "learning_rate": 1.9974323797402264e-05, | |
| "loss": 1.1818, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05439369783363031, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 1.996978978148936e-05, | |
| "loss": 1.0341, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.056269342586514115, | |
| "grad_norm": 7.40625, | |
| "learning_rate": 1.9964887918440735e-05, | |
| "loss": 1.118, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.058144987339397915, | |
| "grad_norm": 7.71875, | |
| "learning_rate": 1.995961838911639e-05, | |
| "loss": 1.2993, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.06002063209228172, | |
| "grad_norm": 7.03125, | |
| "learning_rate": 1.995398138794182e-05, | |
| "loss": 1.1533, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.06189627684516553, | |
| "grad_norm": 7.40625, | |
| "learning_rate": 1.9947977122900825e-05, | |
| "loss": 1.0802, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.06377192159804933, | |
| "grad_norm": 7.15625, | |
| "learning_rate": 1.9941605815527827e-05, | |
| "loss": 1.0135, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.06564756635093313, | |
| "grad_norm": 6.25, | |
| "learning_rate": 1.9934867700899724e-05, | |
| "loss": 1.074, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.06752321110381694, | |
| "grad_norm": 6.9375, | |
| "learning_rate": 1.9927763027627184e-05, | |
| "loss": 1.0574, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06939885585670075, | |
| "grad_norm": 8.4375, | |
| "learning_rate": 1.99202920578455e-05, | |
| "loss": 1.0316, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.07127450060958454, | |
| "grad_norm": 7.28125, | |
| "learning_rate": 1.9912455067204898e-05, | |
| "loss": 1.0653, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07315014536246835, | |
| "grad_norm": 10.4375, | |
| "learning_rate": 1.990425234486038e-05, | |
| "loss": 1.1119, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.07502579011535215, | |
| "grad_norm": 7.71875, | |
| "learning_rate": 1.9895684193461047e-05, | |
| "loss": 1.28, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.07690143486823596, | |
| "grad_norm": 7.6875, | |
| "learning_rate": 1.9886750929138935e-05, | |
| "loss": 1.1392, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.07877707962111977, | |
| "grad_norm": 5.90625, | |
| "learning_rate": 1.987745288149735e-05, | |
| "loss": 1.1955, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.08065272437400356, | |
| "grad_norm": 6.625, | |
| "learning_rate": 1.986779039359871e-05, | |
| "loss": 1.1513, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.08252836912688737, | |
| "grad_norm": 7.75, | |
| "learning_rate": 1.985776382195189e-05, | |
| "loss": 1.0088, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.08440401387977117, | |
| "grad_norm": 7.5625, | |
| "learning_rate": 1.984737353649906e-05, | |
| "loss": 0.9908, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.08627965863265498, | |
| "grad_norm": 8.5, | |
| "learning_rate": 1.9836619920602032e-05, | |
| "loss": 1.1958, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.08815530338553879, | |
| "grad_norm": 11.3125, | |
| "learning_rate": 1.9825503371028136e-05, | |
| "loss": 1.1107, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.09003094813842258, | |
| "grad_norm": 6.78125, | |
| "learning_rate": 1.981402429793556e-05, | |
| "loss": 1.062, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.09190659289130639, | |
| "grad_norm": 7.75, | |
| "learning_rate": 1.980218312485822e-05, | |
| "loss": 1.1393, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.09378223764419019, | |
| "grad_norm": 9.5625, | |
| "learning_rate": 1.978998028869015e-05, | |
| "loss": 1.0907, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.095657882397074, | |
| "grad_norm": 9.4375, | |
| "learning_rate": 1.977741623966936e-05, | |
| "loss": 1.0887, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.0975335271499578, | |
| "grad_norm": 8.125, | |
| "learning_rate": 1.9764491441361227e-05, | |
| "loss": 1.1234, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.0994091719028416, | |
| "grad_norm": 7.03125, | |
| "learning_rate": 1.975120637064142e-05, | |
| "loss": 1.0825, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.1012848166557254, | |
| "grad_norm": 7.5, | |
| "learning_rate": 1.973756151767826e-05, | |
| "loss": 1.0728, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.10316046140860921, | |
| "grad_norm": 7.65625, | |
| "learning_rate": 1.972355738591467e-05, | |
| "loss": 1.2317, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.10503610616149302, | |
| "grad_norm": 8.5625, | |
| "learning_rate": 1.9709194492049585e-05, | |
| "loss": 1.1281, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.10691175091437681, | |
| "grad_norm": 5.84375, | |
| "learning_rate": 1.9694473366018887e-05, | |
| "loss": 0.9848, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.10878739566726062, | |
| "grad_norm": 8.4375, | |
| "learning_rate": 1.9679394550975864e-05, | |
| "loss": 1.0793, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.11066304042014442, | |
| "grad_norm": 8.875, | |
| "learning_rate": 1.9663958603271148e-05, | |
| "loss": 1.1765, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.11253868517302823, | |
| "grad_norm": 6.4375, | |
| "learning_rate": 1.9648166092432216e-05, | |
| "loss": 1.0936, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.11441432992591204, | |
| "grad_norm": 7.21875, | |
| "learning_rate": 1.9632017601142353e-05, | |
| "loss": 1.0343, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.11628997467879583, | |
| "grad_norm": 8.75, | |
| "learning_rate": 1.961551372521916e-05, | |
| "loss": 1.2004, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.11816561943167964, | |
| "grad_norm": 8.5, | |
| "learning_rate": 1.9598655073592583e-05, | |
| "loss": 1.1076, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.12004126418456344, | |
| "grad_norm": 7.0625, | |
| "learning_rate": 1.9581442268282426e-05, | |
| "loss": 1.1502, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.12191690893744725, | |
| "grad_norm": 7.65625, | |
| "learning_rate": 1.956387594437541e-05, | |
| "loss": 1.0411, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.12379255369033106, | |
| "grad_norm": 7.59375, | |
| "learning_rate": 1.9545956750001744e-05, | |
| "loss": 0.9873, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.12566819844321486, | |
| "grad_norm": 6.1875, | |
| "learning_rate": 1.952768534631121e-05, | |
| "loss": 1.1338, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.12754384319609866, | |
| "grad_norm": 9.125, | |
| "learning_rate": 1.950906240744877e-05, | |
| "loss": 1.0202, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.12941948794898248, | |
| "grad_norm": 8.5625, | |
| "learning_rate": 1.9490088620529678e-05, | |
| "loss": 1.1292, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.13129513270186627, | |
| "grad_norm": 5.8125, | |
| "learning_rate": 1.9470764685614158e-05, | |
| "loss": 1.0699, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.13317077745475006, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 1.945109131568154e-05, | |
| "loss": 1.1161, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.13504642220763388, | |
| "grad_norm": 6.125, | |
| "learning_rate": 1.943106923660398e-05, | |
| "loss": 1.099, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.13692206696051767, | |
| "grad_norm": 6.53125, | |
| "learning_rate": 1.9410699187119662e-05, | |
| "loss": 1.0477, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.1387977117134015, | |
| "grad_norm": 8.9375, | |
| "learning_rate": 1.938998191880556e-05, | |
| "loss": 1.0746, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.1406733564662853, | |
| "grad_norm": 7.09375, | |
| "learning_rate": 1.936891819604968e-05, | |
| "loss": 1.0619, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.14254900121916908, | |
| "grad_norm": 6.6875, | |
| "learning_rate": 1.9347508796022888e-05, | |
| "loss": 1.1807, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.1444246459720529, | |
| "grad_norm": 6.75, | |
| "learning_rate": 1.9325754508650208e-05, | |
| "loss": 1.0707, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.1463002907249367, | |
| "grad_norm": 7.1875, | |
| "learning_rate": 1.9303656136581694e-05, | |
| "loss": 1.0464, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.14817593547782051, | |
| "grad_norm": 9.0, | |
| "learning_rate": 1.928121449516281e-05, | |
| "loss": 1.0882, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.1500515802307043, | |
| "grad_norm": 7.71875, | |
| "learning_rate": 1.9258430412404344e-05, | |
| "loss": 1.125, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1519272249835881, | |
| "grad_norm": 6.6875, | |
| "learning_rate": 1.9235304728951868e-05, | |
| "loss": 1.1356, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.15380286973647192, | |
| "grad_norm": 9.9375, | |
| "learning_rate": 1.9211838298054704e-05, | |
| "loss": 1.1532, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.1556785144893557, | |
| "grad_norm": 8.5, | |
| "learning_rate": 1.918803198553446e-05, | |
| "loss": 1.1386, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.15755415924223953, | |
| "grad_norm": 7.96875, | |
| "learning_rate": 1.916388666975307e-05, | |
| "loss": 1.064, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.15942980399512333, | |
| "grad_norm": 9.0, | |
| "learning_rate": 1.9139403241580403e-05, | |
| "loss": 1.2345, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.16130544874800712, | |
| "grad_norm": 7.25, | |
| "learning_rate": 1.9114582604361368e-05, | |
| "loss": 1.1021, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.16318109350089094, | |
| "grad_norm": 9.4375, | |
| "learning_rate": 1.9089425673882617e-05, | |
| "loss": 1.1105, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.16505673825377473, | |
| "grad_norm": 8.375, | |
| "learning_rate": 1.906393337833872e-05, | |
| "loss": 1.0213, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.16693238300665855, | |
| "grad_norm": 6.5625, | |
| "learning_rate": 1.9038106658297946e-05, | |
| "loss": 1.2062, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.16880802775954235, | |
| "grad_norm": 7.6875, | |
| "learning_rate": 1.9011946466667553e-05, | |
| "loss": 1.2408, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.17068367251242614, | |
| "grad_norm": 6.3125, | |
| "learning_rate": 1.8985453768658613e-05, | |
| "loss": 1.0678, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.17255931726530996, | |
| "grad_norm": 7.5, | |
| "learning_rate": 1.8958629541750422e-05, | |
| "loss": 1.071, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.17443496201819375, | |
| "grad_norm": 8.4375, | |
| "learning_rate": 1.893147477565443e-05, | |
| "loss": 1.2118, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.17631060677107757, | |
| "grad_norm": 7.90625, | |
| "learning_rate": 1.8903990472277707e-05, | |
| "loss": 1.0279, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.17818625152396136, | |
| "grad_norm": 8.9375, | |
| "learning_rate": 1.8876177645685997e-05, | |
| "loss": 1.1233, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.18006189627684516, | |
| "grad_norm": 6.78125, | |
| "learning_rate": 1.8848037322066295e-05, | |
| "loss": 1.0398, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.18193754102972898, | |
| "grad_norm": 7.4375, | |
| "learning_rate": 1.881957053968898e-05, | |
| "loss": 1.1796, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.18381318578261277, | |
| "grad_norm": 7.15625, | |
| "learning_rate": 1.8790778348869516e-05, | |
| "loss": 1.0767, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.1856888305354966, | |
| "grad_norm": 8.375, | |
| "learning_rate": 1.8761661811929686e-05, | |
| "loss": 1.1285, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.18756447528838038, | |
| "grad_norm": 8.875, | |
| "learning_rate": 1.8732222003158423e-05, | |
| "loss": 1.163, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.18756447528838038, | |
| "eval_loss": 1.1088175773620605, | |
| "eval_runtime": 108.1214, | |
| "eval_samples_per_second": 20.486, | |
| "eval_steps_per_second": 2.562, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.18944012004126418, | |
| "grad_norm": 8.0625, | |
| "learning_rate": 1.870246000877214e-05, | |
| "loss": 1.031, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.191315764794148, | |
| "grad_norm": 6.84375, | |
| "learning_rate": 1.8672376926874668e-05, | |
| "loss": 1.1465, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.1931914095470318, | |
| "grad_norm": 6.90625, | |
| "learning_rate": 1.8641973867416742e-05, | |
| "loss": 1.1154, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.1950670542999156, | |
| "grad_norm": 7.375, | |
| "learning_rate": 1.8611251952155057e-05, | |
| "loss": 1.0743, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.1969426990527994, | |
| "grad_norm": 7.40625, | |
| "learning_rate": 1.8580212314610847e-05, | |
| "loss": 1.0347, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.1988183438056832, | |
| "grad_norm": 6.28125, | |
| "learning_rate": 1.85488561000281e-05, | |
| "loss": 1.0763, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.20069398855856702, | |
| "grad_norm": 5.84375, | |
| "learning_rate": 1.8517184465331288e-05, | |
| "loss": 1.078, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.2025696333114508, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 1.848519857908267e-05, | |
| "loss": 1.178, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.2044452780643346, | |
| "grad_norm": 7.375, | |
| "learning_rate": 1.845289962143918e-05, | |
| "loss": 1.0804, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.20632092281721842, | |
| "grad_norm": 7.96875, | |
| "learning_rate": 1.8420288784108917e-05, | |
| "loss": 1.1589, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.20819656757010221, | |
| "grad_norm": 7.375, | |
| "learning_rate": 1.8387367270307122e-05, | |
| "loss": 1.1116, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.21007221232298603, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 1.835413629471182e-05, | |
| "loss": 0.9735, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.21194785707586983, | |
| "grad_norm": 7.625, | |
| "learning_rate": 1.832059708341899e-05, | |
| "loss": 1.0991, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.21382350182875362, | |
| "grad_norm": 9.125, | |
| "learning_rate": 1.8286750873897338e-05, | |
| "loss": 1.1227, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.21569914658163744, | |
| "grad_norm": 10.0, | |
| "learning_rate": 1.8252598914942624e-05, | |
| "loss": 0.9956, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.21757479133452123, | |
| "grad_norm": 7.8125, | |
| "learning_rate": 1.8218142466631595e-05, | |
| "loss": 1.0511, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.21945043608740505, | |
| "grad_norm": 7.0, | |
| "learning_rate": 1.8183382800275492e-05, | |
| "loss": 1.1195, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.22132608084028885, | |
| "grad_norm": 8.125, | |
| "learning_rate": 1.8148321198373146e-05, | |
| "loss": 1.0831, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.22320172559317264, | |
| "grad_norm": 9.0625, | |
| "learning_rate": 1.8112958954563647e-05, | |
| "loss": 1.2181, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.22507737034605646, | |
| "grad_norm": 5.6875, | |
| "learning_rate": 1.8077297373578625e-05, | |
| "loss": 1.0144, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.22695301509894025, | |
| "grad_norm": 7.0, | |
| "learning_rate": 1.8041337771194124e-05, | |
| "loss": 0.8084, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.22882865985182407, | |
| "grad_norm": 6.46875, | |
| "learning_rate": 1.800508147418201e-05, | |
| "loss": 1.0292, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.23070430460470787, | |
| "grad_norm": 7.25, | |
| "learning_rate": 1.796852982026107e-05, | |
| "loss": 1.0705, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.23257994935759166, | |
| "grad_norm": 9.4375, | |
| "learning_rate": 1.7931684158047623e-05, | |
| "loss": 1.0863, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.23445559411047548, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 1.7894545847005764e-05, | |
| "loss": 1.1707, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.23633123886335927, | |
| "grad_norm": 7.71875, | |
| "learning_rate": 1.7857116257397225e-05, | |
| "loss": 0.9151, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.2382068836162431, | |
| "grad_norm": 7.5625, | |
| "learning_rate": 1.7819396770230796e-05, | |
| "loss": 1.0119, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.24008252836912689, | |
| "grad_norm": 7.21875, | |
| "learning_rate": 1.7781388777211374e-05, | |
| "loss": 1.1517, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.24195817312201068, | |
| "grad_norm": 7.96875, | |
| "learning_rate": 1.7743093680688626e-05, | |
| "loss": 1.0162, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.2438338178748945, | |
| "grad_norm": 7.78125, | |
| "learning_rate": 1.7704512893605247e-05, | |
| "loss": 1.208, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2457094626277783, | |
| "grad_norm": 6.34375, | |
| "learning_rate": 1.7665647839444807e-05, | |
| "loss": 1.0252, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.2475851073806621, | |
| "grad_norm": 7.90625, | |
| "learning_rate": 1.7626499952179255e-05, | |
| "loss": 1.1601, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2494607521335459, | |
| "grad_norm": 8.625, | |
| "learning_rate": 1.7587070676215995e-05, | |
| "loss": 1.1795, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.2513363968864297, | |
| "grad_norm": 6.59375, | |
| "learning_rate": 1.75473614663446e-05, | |
| "loss": 0.9092, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.2532120416393135, | |
| "grad_norm": 12.5625, | |
| "learning_rate": 1.750737378768314e-05, | |
| "loss": 1.1697, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.2550876863921973, | |
| "grad_norm": 7.59375, | |
| "learning_rate": 1.7467109115624113e-05, | |
| "loss": 1.1992, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.25696333114508113, | |
| "grad_norm": 7.4375, | |
| "learning_rate": 1.7426568935780007e-05, | |
| "loss": 1.0365, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.25883897589796495, | |
| "grad_norm": 7.84375, | |
| "learning_rate": 1.7385754743928512e-05, | |
| "loss": 1.2158, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.2607146206508487, | |
| "grad_norm": 7.5, | |
| "learning_rate": 1.7344668045957303e-05, | |
| "loss": 1.1652, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.26259026540373254, | |
| "grad_norm": 7.1875, | |
| "learning_rate": 1.730331035780849e-05, | |
| "loss": 1.1178, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.26446591015661636, | |
| "grad_norm": 10.375, | |
| "learning_rate": 1.726168320542269e-05, | |
| "loss": 1.0897, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.2663415549095001, | |
| "grad_norm": 7.625, | |
| "learning_rate": 1.7219788124682702e-05, | |
| "loss": 1.2076, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.26821719966238394, | |
| "grad_norm": 6.15625, | |
| "learning_rate": 1.7177626661356885e-05, | |
| "loss": 1.0381, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.27009284441526776, | |
| "grad_norm": 8.25, | |
| "learning_rate": 1.713520037104208e-05, | |
| "loss": 1.0232, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.27196848916815153, | |
| "grad_norm": 7.4375, | |
| "learning_rate": 1.709251081910623e-05, | |
| "loss": 1.0218, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.27384413392103535, | |
| "grad_norm": 9.625, | |
| "learning_rate": 1.704955958063063e-05, | |
| "loss": 1.0518, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.27571977867391917, | |
| "grad_norm": 8.3125, | |
| "learning_rate": 1.700634824035182e-05, | |
| "loss": 1.043, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.277595423426803, | |
| "grad_norm": 7.0, | |
| "learning_rate": 1.696287839260308e-05, | |
| "loss": 1.0724, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.27947106817968675, | |
| "grad_norm": 5.84375, | |
| "learning_rate": 1.6919151641255642e-05, | |
| "loss": 0.9938, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.2813467129325706, | |
| "grad_norm": 9.4375, | |
| "learning_rate": 1.6875169599659495e-05, | |
| "loss": 1.1153, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.2832223576854544, | |
| "grad_norm": 7.375, | |
| "learning_rate": 1.6830933890583863e-05, | |
| "loss": 1.1801, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.28509800243833816, | |
| "grad_norm": 7.4375, | |
| "learning_rate": 1.6786446146157332e-05, | |
| "loss": 1.1175, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.286973647191222, | |
| "grad_norm": 8.0, | |
| "learning_rate": 1.6741708007807626e-05, | |
| "loss": 1.1015, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.2888492919441058, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 1.6696721126201048e-05, | |
| "loss": 1.2244, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.29072493669698957, | |
| "grad_norm": 9.5625, | |
| "learning_rate": 1.6651487161181577e-05, | |
| "loss": 1.0448, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.2926005814498734, | |
| "grad_norm": 8.6875, | |
| "learning_rate": 1.6606007781709626e-05, | |
| "loss": 1.0295, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.2944762262027572, | |
| "grad_norm": 5.625, | |
| "learning_rate": 1.6560284665800464e-05, | |
| "loss": 0.9881, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.29635187095564103, | |
| "grad_norm": 10.0625, | |
| "learning_rate": 1.6514319500462303e-05, | |
| "loss": 1.0681, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.2982275157085248, | |
| "grad_norm": 6.71875, | |
| "learning_rate": 1.646811398163405e-05, | |
| "loss": 1.0571, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.3001031604614086, | |
| "grad_norm": 5.6875, | |
| "learning_rate": 1.642166981412274e-05, | |
| "loss": 0.9326, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.30197880521429243, | |
| "grad_norm": 6.5625, | |
| "learning_rate": 1.6374988711540634e-05, | |
| "loss": 1.2067, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.3038544499671762, | |
| "grad_norm": 7.75, | |
| "learning_rate": 1.6328072396241993e-05, | |
| "loss": 1.1231, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.30573009472006, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 1.6280922599259515e-05, | |
| "loss": 1.2164, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.30760573947294384, | |
| "grad_norm": 9.5625, | |
| "learning_rate": 1.62335410602405e-05, | |
| "loss": 1.0431, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.3094813842258276, | |
| "grad_norm": 9.3125, | |
| "learning_rate": 1.6185929527382628e-05, | |
| "loss": 1.2427, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.3113570289787114, | |
| "grad_norm": 7.125, | |
| "learning_rate": 1.6138089757369475e-05, | |
| "loss": 1.1003, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.31323267373159525, | |
| "grad_norm": 5.78125, | |
| "learning_rate": 1.6090023515305703e-05, | |
| "loss": 1.1261, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.31510831848447907, | |
| "grad_norm": 5.96875, | |
| "learning_rate": 1.604173257465192e-05, | |
| "loss": 0.9375, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.31698396323736283, | |
| "grad_norm": 7.875, | |
| "learning_rate": 1.5993218717159253e-05, | |
| "loss": 1.0321, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.31885960799024665, | |
| "grad_norm": 8.0625, | |
| "learning_rate": 1.5944483732803612e-05, | |
| "loss": 1.1149, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.3207352527431305, | |
| "grad_norm": 10.125, | |
| "learning_rate": 1.5895529419719645e-05, | |
| "loss": 0.8962, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.32261089749601424, | |
| "grad_norm": 7.5, | |
| "learning_rate": 1.5846357584134385e-05, | |
| "loss": 0.9511, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.32448654224889806, | |
| "grad_norm": 6.375, | |
| "learning_rate": 1.579697004030061e-05, | |
| "loss": 1.0865, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.3263621870017819, | |
| "grad_norm": 8.0625, | |
| "learning_rate": 1.5747368610429933e-05, | |
| "loss": 0.9981, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.32823783175466564, | |
| "grad_norm": 6.90625, | |
| "learning_rate": 1.569755512462551e-05, | |
| "loss": 1.1338, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.33011347650754946, | |
| "grad_norm": 6.6875, | |
| "learning_rate": 1.5647531420814574e-05, | |
| "loss": 0.9559, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.3319891212604333, | |
| "grad_norm": 7.59375, | |
| "learning_rate": 1.559729934468059e-05, | |
| "loss": 0.949, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.3338647660133171, | |
| "grad_norm": 8.75, | |
| "learning_rate": 1.5546860749595165e-05, | |
| "loss": 1.0738, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.33574041076620087, | |
| "grad_norm": 10.6875, | |
| "learning_rate": 1.5496217496549673e-05, | |
| "loss": 1.2675, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.3376160555190847, | |
| "grad_norm": 8.125, | |
| "learning_rate": 1.5445371454086574e-05, | |
| "loss": 1.1757, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.3394917002719685, | |
| "grad_norm": 7.59375, | |
| "learning_rate": 1.5394324498230487e-05, | |
| "loss": 1.1336, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.3413673450248523, | |
| "grad_norm": 7.65625, | |
| "learning_rate": 1.5343078512418977e-05, | |
| "loss": 1.008, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.3432429897777361, | |
| "grad_norm": 7.40625, | |
| "learning_rate": 1.529163538743303e-05, | |
| "loss": 1.1445, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.3451186345306199, | |
| "grad_norm": 6.5, | |
| "learning_rate": 1.5239997021327343e-05, | |
| "loss": 1.0209, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.3469942792835037, | |
| "grad_norm": 6.59375, | |
| "learning_rate": 1.518816531936024e-05, | |
| "loss": 1.0617, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.3488699240363875, | |
| "grad_norm": 7.90625, | |
| "learning_rate": 1.5136142193923413e-05, | |
| "loss": 1.0885, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.3507455687892713, | |
| "grad_norm": 7.875, | |
| "learning_rate": 1.5083929564471344e-05, | |
| "loss": 0.9534, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.35262121354215514, | |
| "grad_norm": 6.8125, | |
| "learning_rate": 1.5031529357450487e-05, | |
| "loss": 0.9802, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.3544968582950389, | |
| "grad_norm": 11.375, | |
| "learning_rate": 1.4978943506228198e-05, | |
| "loss": 1.149, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.35637250304792273, | |
| "grad_norm": 6.25, | |
| "learning_rate": 1.4926173951021384e-05, | |
| "loss": 1.179, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.35824814780080655, | |
| "grad_norm": 9.0625, | |
| "learning_rate": 1.4873222638824938e-05, | |
| "loss": 0.9677, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.3601237925536903, | |
| "grad_norm": 7.75, | |
| "learning_rate": 1.4820091523339883e-05, | |
| "loss": 1.0542, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.36199943730657413, | |
| "grad_norm": 7.0, | |
| "learning_rate": 1.4766782564901299e-05, | |
| "loss": 0.9912, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.36387508205945795, | |
| "grad_norm": 8.5, | |
| "learning_rate": 1.471329773040599e-05, | |
| "loss": 1.1461, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.3657507268123417, | |
| "grad_norm": 9.25, | |
| "learning_rate": 1.465963899323992e-05, | |
| "loss": 1.1274, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.36762637156522554, | |
| "grad_norm": 7.03125, | |
| "learning_rate": 1.4605808333205387e-05, | |
| "loss": 1.0867, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.36950201631810936, | |
| "grad_norm": 8.875, | |
| "learning_rate": 1.4551807736447996e-05, | |
| "loss": 1.0081, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.3713776610709932, | |
| "grad_norm": 6.96875, | |
| "learning_rate": 1.4497639195383362e-05, | |
| "loss": 0.9978, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.37325330582387695, | |
| "grad_norm": 7.875, | |
| "learning_rate": 1.4443304708623598e-05, | |
| "loss": 0.9439, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.37512895057676077, | |
| "grad_norm": 7.1875, | |
| "learning_rate": 1.4388806280903591e-05, | |
| "loss": 0.9749, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.37512895057676077, | |
| "eval_loss": 1.0913933515548706, | |
| "eval_runtime": 107.8156, | |
| "eval_samples_per_second": 20.544, | |
| "eval_steps_per_second": 2.569, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 2666, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 3, | |
| "early_stopping_threshold": 0.001 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.3299712794624e+16, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |