| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.6506710044733631, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.999967339434433e-05, | |
| "loss": 1.343, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9998820960283476e-05, | |
| "loss": 1.1858, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.99974394497948e-05, | |
| "loss": 1.1057, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9995528892116685e-05, | |
| "loss": 1.0698, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999308932768425e-05, | |
| "loss": 1.0316, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.99897583236926e-05, | |
| "loss": 1.0046, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9985774375988764e-05, | |
| "loss": 0.9848, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.998113758866712e-05, | |
| "loss": 0.9656, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.997584808287975e-05, | |
| "loss": 0.9648, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.996990599683325e-05, | |
| "loss": 0.9415, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.996331148578514e-05, | |
| "loss": 0.9388, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.995606472203979e-05, | |
| "loss": 0.9289, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.994816589494391e-05, | |
| "loss": 0.9252, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.993961521088163e-05, | |
| "loss": 0.91, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9930412893269106e-05, | |
| "loss": 0.9033, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.992055918254865e-05, | |
| "loss": 0.9064, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9910054336182467e-05, | |
| "loss": 0.8946, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.989889862864594e-05, | |
| "loss": 0.8914, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.988709235142044e-05, | |
| "loss": 0.9007, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9874635812985734e-05, | |
| "loss": 0.9008, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9861529338811874e-05, | |
| "loss": 0.8714, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.984777327135076e-05, | |
| "loss": 0.8865, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9833367970027145e-05, | |
| "loss": 0.8716, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.981831381122926e-05, | |
| "loss": 0.8817, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.980261118829896e-05, | |
| "loss": 0.8658, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.978626051152152e-05, | |
| "loss": 0.8722, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.976926220811479e-05, | |
| "loss": 0.8541, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.975161672221813e-05, | |
| "loss": 0.8504, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.97333245148808e-05, | |
| "loss": 0.8467, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.971438606404986e-05, | |
| "loss": 0.8533, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.969480186455771e-05, | |
| "loss": 0.853, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9674572428109205e-05, | |
| "loss": 0.8432, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9653698283268183e-05, | |
| "loss": 0.854, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.963217997544376e-05, | |
| "loss": 0.8435, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.961001806687602e-05, | |
| "loss": 0.8518, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.958721313662133e-05, | |
| "loss": 0.848, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.956376578053723e-05, | |
| "loss": 0.8436, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9539676611266846e-05, | |
| "loss": 0.8405, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9514946258222895e-05, | |
| "loss": 0.838, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.948957536757123e-05, | |
| "loss": 0.8458, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.946356460221395e-05, | |
| "loss": 0.8379, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.943691464177212e-05, | |
| "loss": 0.827, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.940962618256796e-05, | |
| "loss": 0.8208, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9381699937606674e-05, | |
| "loss": 0.8242, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.935313663655782e-05, | |
| "loss": 0.8362, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.932393702573627e-05, | |
| "loss": 0.8363, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.929410186808264e-05, | |
| "loss": 0.8275, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.926363194314344e-05, | |
| "loss": 0.821, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.923252804705068e-05, | |
| "loss": 0.8271, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.920079099250099e-05, | |
| "loss": 0.8178, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.916842160873451e-05, | |
| "loss": 0.8293, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.913542074151315e-05, | |
| "loss": 0.8294, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.910178925309849e-05, | |
| "loss": 0.8202, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.906752802222928e-05, | |
| "loss": 0.8301, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.903263794409846e-05, | |
| "loss": 0.8031, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.8997119930329785e-05, | |
| "loss": 0.821, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.896097490895398e-05, | |
| "loss": 0.8132, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.8924203824384516e-05, | |
| "loss": 0.8113, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.888680763739294e-05, | |
| "loss": 0.8199, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8848787325083734e-05, | |
| "loss": 0.8196, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8810143880868816e-05, | |
| "loss": 0.8128, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8770878314441606e-05, | |
| "loss": 0.8123, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8730991651750565e-05, | |
| "loss": 0.8048, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.869048493497247e-05, | |
| "loss": 0.8119, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.864935922248515e-05, | |
| "loss": 0.8141, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.860761558883981e-05, | |
| "loss": 0.8, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.856525512473302e-05, | |
| "loss": 0.816, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.852227893697814e-05, | |
| "loss": 0.7976, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.847868814847646e-05, | |
| "loss": 0.8121, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.84344838981878e-05, | |
| "loss": 0.8175, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.8389667341100844e-05, | |
| "loss": 0.8109, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.834423964820284e-05, | |
| "loss": 0.7948, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.8298202006449136e-05, | |
| "loss": 0.7986, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.8251555618732036e-05, | |
| "loss": 0.8049, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.820430170384947e-05, | |
| "loss": 0.7981, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.8156441496473126e-05, | |
| "loss": 0.799, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.8107976247116145e-05, | |
| "loss": 0.804, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.805890722210048e-05, | |
| "loss": 0.8053, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.8009235703523824e-05, | |
| "loss": 0.7969, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.79589629892261e-05, | |
| "loss": 0.8051, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.7908090392755526e-05, | |
| "loss": 0.796, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.785661924333431e-05, | |
| "loss": 0.7928, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.780455088582394e-05, | |
| "loss": 0.8017, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.775188668069002e-05, | |
| "loss": 0.7969, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.769862800396674e-05, | |
| "loss": 0.7942, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.764477624722088e-05, | |
| "loss": 0.7877, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.759033281751553e-05, | |
| "loss": 0.7979, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.7535299137373236e-05, | |
| "loss": 0.7986, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.7479676644738894e-05, | |
| "loss": 0.8124, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.742346679294217e-05, | |
| "loss": 0.7959, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.7366671050659505e-05, | |
| "loss": 0.7884, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.730929090187573e-05, | |
| "loss": 0.7911, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.725132784584536e-05, | |
| "loss": 0.7937, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.719278339705333e-05, | |
| "loss": 0.7994, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.71336590851755e-05, | |
| "loss": 0.7905, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.707395645503863e-05, | |
| "loss": 0.7924, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.701367706658006e-05, | |
| "loss": 0.7891, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.695282249480692e-05, | |
| "loss": 0.7962, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.6891394329755e-05, | |
| "loss": 0.7882, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.6829394176447195e-05, | |
| "loss": 0.8011, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.676682365485154e-05, | |
| "loss": 0.7955, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.670368439983896e-05, | |
| "loss": 0.7906, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.663997806114048e-05, | |
| "loss": 0.7853, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.657570630330411e-05, | |
| "loss": 0.7825, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.6510870805651444e-05, | |
| "loss": 0.7768, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.6445473262233696e-05, | |
| "loss": 0.7795, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.637951538178746e-05, | |
| "loss": 0.7832, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.63129988876901e-05, | |
| "loss": 0.7934, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.624592551791465e-05, | |
| "loss": 0.7892, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.617829702498446e-05, | |
| "loss": 0.7819, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.611011517592741e-05, | |
| "loss": 0.7868, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.604138175222968e-05, | |
| "loss": 0.7914, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.597209854978929e-05, | |
| "loss": 0.7817, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.590226737886908e-05, | |
| "loss": 0.7916, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.583189006404949e-05, | |
| "loss": 0.7799, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.576096844418083e-05, | |
| "loss": 0.7776, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.568950437233529e-05, | |
| "loss": 0.7803, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.561749971575846e-05, | |
| "loss": 0.7884, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.5544956355820596e-05, | |
| "loss": 0.7827, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.547187618796742e-05, | |
| "loss": 0.7857, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.5398261121670634e-05, | |
| "loss": 0.7797, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.532411308037798e-05, | |
| "loss": 0.7821, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.524943400146306e-05, | |
| "loss": 0.7747, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.51742258361746e-05, | |
| "loss": 0.7737, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.509849054958559e-05, | |
| "loss": 0.7868, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.502223012054186e-05, | |
| "loss": 0.7818, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.4945446541610414e-05, | |
| "loss": 0.7804, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.486814181902734e-05, | |
| "loss": 0.7759, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.4790317972645396e-05, | |
| "loss": 0.7744, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.4711977035881256e-05, | |
| "loss": 0.7814, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.463312105566237e-05, | |
| "loss": 0.7611, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.4553752092373465e-05, | |
| "loss": 0.7796, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.447387221980272e-05, | |
| "loss": 0.7839, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.4393483525087584e-05, | |
| "loss": 0.7726, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.4312588108660246e-05, | |
| "loss": 0.7763, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.423118808419276e-05, | |
| "loss": 0.766, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.4149285578541785e-05, | |
| "loss": 0.7887, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.406688273169305e-05, | |
| "loss": 0.7637, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.3983981696705415e-05, | |
| "loss": 0.7609, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.390058463965464e-05, | |
| "loss": 0.7729, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.381669373957675e-05, | |
| "loss": 0.7751, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.3732311188411154e-05, | |
| "loss": 0.7877, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.3647439190943315e-05, | |
| "loss": 0.7738, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.3562079964747207e-05, | |
| "loss": 0.7834, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.347623574012728e-05, | |
| "loss": 0.7835, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.338990876006031e-05, | |
| "loss": 0.7707, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.330310128013666e-05, | |
| "loss": 0.7674, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.321581556850146e-05, | |
| "loss": 0.7687, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.312805390579527e-05, | |
| "loss": 0.7663, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.3039818585094524e-05, | |
| "loss": 0.7662, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.29511119118516e-05, | |
| "loss": 0.7688, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.28619362038346e-05, | |
| "loss": 0.7652, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.277229379106676e-05, | |
| "loss": 0.7683, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.2682187015765615e-05, | |
| "loss": 0.7621, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.259161823228174e-05, | |
| "loss": 0.7778, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.2500589807037306e-05, | |
| "loss": 0.7809, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.240910411846418e-05, | |
| "loss": 0.7717, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.231716355694184e-05, | |
| "loss": 0.7584, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.222477052473486e-05, | |
| "loss": 0.7631, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.21319274359302e-05, | |
| "loss": 0.7681, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.203863671637409e-05, | |
| "loss": 0.7739, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.194490080360866e-05, | |
| "loss": 0.7646, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1850722146808254e-05, | |
| "loss": 0.7647, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1756103206715416e-05, | |
| "loss": 0.7675, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.166104645557662e-05, | |
| "loss": 0.7652, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.15655543770777e-05, | |
| "loss": 0.765, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.146962946627886e-05, | |
| "loss": 0.7614, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.137327422954958e-05, | |
| "loss": 0.7544, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.127649118450308e-05, | |
| "loss": 0.7673, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.117928285993055e-05, | |
| "loss": 0.7593, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.108165179573508e-05, | |
| "loss": 0.7735, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.0983600542865285e-05, | |
| "loss": 0.7666, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.088513166324867e-05, | |
| "loss": 0.7627, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.078624772972467e-05, | |
| "loss": 0.77, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.068695132597744e-05, | |
| "loss": 0.7569, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.058724504646834e-05, | |
| "loss": 0.7654, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.048713149636816e-05, | |
| "loss": 0.753, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.038661329148904e-05, | |
| "loss": 0.7705, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.02856930582161e-05, | |
| "loss": 0.7542, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.018437343343886e-05, | |
| "loss": 0.7532, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.008265706448234e-05, | |
| "loss": 0.7662, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.9980546609037825e-05, | |
| "loss": 0.7607, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.9878044735093505e-05, | |
| "loss": 0.7552, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.977515412086472e-05, | |
| "loss": 0.7645, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.9671877454723985e-05, | |
| "loss": 0.7689, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.956821743513078e-05, | |
| "loss": 0.7615, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.946417677056097e-05, | |
| "loss": 0.7703, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.935975817943614e-05, | |
| "loss": 0.766, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.925496439005246e-05, | |
| "loss": 0.7632, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.914979814050949e-05, | |
| "loss": 0.7646, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.904426217863858e-05, | |
| "loss": 0.7624, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.8938359261931076e-05, | |
| "loss": 0.7498, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.883209215746633e-05, | |
| "loss": 0.7541, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.8725463641839296e-05, | |
| "loss": 0.7622, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.8618476501088094e-05, | |
| "loss": 0.7536, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.851113353062115e-05, | |
| "loss": 0.7419, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.840343753514414e-05, | |
| "loss": 0.756, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.829539132858678e-05, | |
| "loss": 0.7511, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.818699773402925e-05, | |
| "loss": 0.7505, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.80782595836284e-05, | |
| "loss": 0.7396, | |
| "step": 2000 | |
| } | |
| ], | |
| "max_steps": 6146, | |
| "num_train_epochs": 2, | |
| "total_flos": 3.592019131911635e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |