| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9992697108389406, | |
| "eval_steps": 500, | |
| "global_step": 358500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.993031591974624e-05, | |
| "loss": 0.6149, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.986063183949248e-05, | |
| "loss": 0.4578, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9790947759238714e-05, | |
| "loss": 0.4321, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.972126367898496e-05, | |
| "loss": 0.4103, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9651579598731195e-05, | |
| "loss": 0.4003, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.958189551847743e-05, | |
| "loss": 0.3793, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.951221143822367e-05, | |
| "loss": 0.3678, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.944252735796991e-05, | |
| "loss": 0.386, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9372843277716144e-05, | |
| "loss": 0.3704, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.930315919746239e-05, | |
| "loss": 0.3623, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9233475117208625e-05, | |
| "loss": 0.3508, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.916379103695486e-05, | |
| "loss": 0.3505, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.90941069567011e-05, | |
| "loss": 0.3406, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.902442287644734e-05, | |
| "loss": 0.337, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.895473879619358e-05, | |
| "loss": 0.342, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.888505471593982e-05, | |
| "loss": 0.3322, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.881537063568606e-05, | |
| "loss": 0.3352, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.87456865554323e-05, | |
| "loss": 0.3293, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.8676002475178536e-05, | |
| "loss": 0.3249, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.860631839492477e-05, | |
| "loss": 0.3265, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.853663431467101e-05, | |
| "loss": 0.3125, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.846695023441725e-05, | |
| "loss": 0.3207, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.8397266154163485e-05, | |
| "loss": 0.3271, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.832758207390973e-05, | |
| "loss": 0.3187, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.8257897993655966e-05, | |
| "loss": 0.3163, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.81882139134022e-05, | |
| "loss": 0.3197, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.811852983314844e-05, | |
| "loss": 0.3132, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.804884575289468e-05, | |
| "loss": 0.306, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.7979161672640915e-05, | |
| "loss": 0.3053, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.790947759238716e-05, | |
| "loss": 0.3011, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.7839793512133396e-05, | |
| "loss": 0.3033, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.777010943187963e-05, | |
| "loss": 0.3103, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.770042535162587e-05, | |
| "loss": 0.3059, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.763074127137211e-05, | |
| "loss": 0.298, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.7561057191118345e-05, | |
| "loss": 0.2967, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.749137311086458e-05, | |
| "loss": 0.2844, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.7421689030610826e-05, | |
| "loss": 0.3001, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.735200495035706e-05, | |
| "loss": 0.3088, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.72823208701033e-05, | |
| "loss": 0.3079, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.721263678984954e-05, | |
| "loss": 0.2908, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.714295270959578e-05, | |
| "loss": 0.2879, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.707326862934202e-05, | |
| "loss": 0.2891, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.7003584549088256e-05, | |
| "loss": 0.3005, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.69339004688345e-05, | |
| "loss": 0.3008, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.686421638858074e-05, | |
| "loss": 0.2914, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.6794532308326974e-05, | |
| "loss": 0.2864, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.672484822807321e-05, | |
| "loss": 0.2885, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.665516414781945e-05, | |
| "loss": 0.289, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.6585480067565685e-05, | |
| "loss": 0.283, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.651579598731193e-05, | |
| "loss": 0.2917, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.6446111907058167e-05, | |
| "loss": 0.2895, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.6376427826804404e-05, | |
| "loss": 0.276, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.630674374655064e-05, | |
| "loss": 0.2847, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.623705966629688e-05, | |
| "loss": 0.2774, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.6167375586043115e-05, | |
| "loss": 0.283, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.609769150578935e-05, | |
| "loss": 0.2859, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.6028007425535596e-05, | |
| "loss": 0.2793, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.5958323345281834e-05, | |
| "loss": 0.2769, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.588863926502807e-05, | |
| "loss": 0.2763, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.581895518477431e-05, | |
| "loss": 0.2886, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5749271104520545e-05, | |
| "loss": 0.2721, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.567958702426678e-05, | |
| "loss": 0.2805, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.560990294401302e-05, | |
| "loss": 0.2779, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5540218863759264e-05, | |
| "loss": 0.2833, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.54705347835055e-05, | |
| "loss": 0.2881, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.540085070325174e-05, | |
| "loss": 0.2802, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5331166622997975e-05, | |
| "loss": 0.2683, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.526148254274422e-05, | |
| "loss": 0.2732, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.5191798462490456e-05, | |
| "loss": 0.2773, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.5122114382236693e-05, | |
| "loss": 0.2791, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.505243030198294e-05, | |
| "loss": 0.2842, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.4982746221729175e-05, | |
| "loss": 0.278, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.491306214147541e-05, | |
| "loss": 0.2757, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.484337806122165e-05, | |
| "loss": 0.2741, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.4773693980967886e-05, | |
| "loss": 0.2708, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.470400990071412e-05, | |
| "loss": 0.2725, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.463432582046037e-05, | |
| "loss": 0.2735, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.4564641740206604e-05, | |
| "loss": 0.2735, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.449495765995284e-05, | |
| "loss": 0.2704, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.442527357969908e-05, | |
| "loss": 0.2757, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.4355589499445316e-05, | |
| "loss": 0.2851, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.428590541919155e-05, | |
| "loss": 0.2681, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.421622133893779e-05, | |
| "loss": 0.27, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.4146537258684034e-05, | |
| "loss": 0.2701, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.407685317843027e-05, | |
| "loss": 0.2648, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.400716909817651e-05, | |
| "loss": 0.2698, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.3937485017922746e-05, | |
| "loss": 0.2693, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.386780093766898e-05, | |
| "loss": 0.2665, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.379811685741522e-05, | |
| "loss": 0.2688, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.3728432777161464e-05, | |
| "loss": 0.2689, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.36587486969077e-05, | |
| "loss": 0.273, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.358906461665394e-05, | |
| "loss": 0.2717, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.3519380536400176e-05, | |
| "loss": 0.2691, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.344969645614641e-05, | |
| "loss": 0.2596, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.338001237589266e-05, | |
| "loss": 0.2639, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.3310328295638894e-05, | |
| "loss": 0.2724, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.324064421538513e-05, | |
| "loss": 0.2572, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.3170960135131375e-05, | |
| "loss": 0.2657, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.310127605487761e-05, | |
| "loss": 0.2681, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.303159197462385e-05, | |
| "loss": 0.2654, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.296190789437009e-05, | |
| "loss": 0.2655, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.2892223814116324e-05, | |
| "loss": 0.2658, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.282253973386256e-05, | |
| "loss": 0.2582, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.2752855653608805e-05, | |
| "loss": 0.2664, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.268317157335504e-05, | |
| "loss": 0.2508, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.261348749310128e-05, | |
| "loss": 0.2569, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.254380341284752e-05, | |
| "loss": 0.2592, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.2474119332593754e-05, | |
| "loss": 0.2645, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.240443525233999e-05, | |
| "loss": 0.2597, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.2334751172086235e-05, | |
| "loss": 0.2545, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.226506709183247e-05, | |
| "loss": 0.2652, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.219538301157871e-05, | |
| "loss": 0.2677, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.2125698931324947e-05, | |
| "loss": 0.2671, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.2056014851071184e-05, | |
| "loss": 0.2601, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.198633077081742e-05, | |
| "loss": 0.2728, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.191664669056366e-05, | |
| "loss": 0.2628, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.18469626103099e-05, | |
| "loss": 0.2628, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.177727853005614e-05, | |
| "loss": 0.2508, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.1707594449802376e-05, | |
| "loss": 0.2661, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.1637910369548614e-05, | |
| "loss": 0.2575, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.156822628929485e-05, | |
| "loss": 0.2537, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.149854220904109e-05, | |
| "loss": 0.2495, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.142885812878733e-05, | |
| "loss": 0.2564, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.135917404853357e-05, | |
| "loss": 0.2515, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.128948996827981e-05, | |
| "loss": 0.2582, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.121980588802605e-05, | |
| "loss": 0.266, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.115012180777229e-05, | |
| "loss": 0.2492, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.1080437727518525e-05, | |
| "loss": 0.2677, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.101075364726476e-05, | |
| "loss": 0.257, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.0941069567011006e-05, | |
| "loss": 0.2586, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.087138548675724e-05, | |
| "loss": 0.2571, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.080170140650348e-05, | |
| "loss": 0.2549, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.073201732624972e-05, | |
| "loss": 0.258, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.0662333245995955e-05, | |
| "loss": 0.2594, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.059264916574219e-05, | |
| "loss": 0.2564, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.052296508548843e-05, | |
| "loss": 0.2646, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.045328100523467e-05, | |
| "loss": 0.2577, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.038359692498091e-05, | |
| "loss": 0.2567, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.031391284472715e-05, | |
| "loss": 0.2518, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.0244228764473384e-05, | |
| "loss": 0.2541, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.017454468421962e-05, | |
| "loss": 0.2504, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.010486060396586e-05, | |
| "loss": 0.2595, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.00351765237121e-05, | |
| "loss": 0.2618, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.996549244345834e-05, | |
| "loss": 0.2604, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.989580836320458e-05, | |
| "loss": 0.2525, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.9826124282950814e-05, | |
| "loss": 0.2447, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.975644020269705e-05, | |
| "loss": 0.2535, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.968675612244329e-05, | |
| "loss": 0.256, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.9617072042189526e-05, | |
| "loss": 0.2475, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.954738796193577e-05, | |
| "loss": 0.2591, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.947770388168201e-05, | |
| "loss": 0.2464, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.9408019801428244e-05, | |
| "loss": 0.2535, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.933833572117449e-05, | |
| "loss": 0.2514, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.9268651640920725e-05, | |
| "loss": 0.2484, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.919896756066696e-05, | |
| "loss": 0.2483, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.91292834804132e-05, | |
| "loss": 0.2559, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.9059599400159444e-05, | |
| "loss": 0.2526, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.898991531990568e-05, | |
| "loss": 0.2537, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.892023123965192e-05, | |
| "loss": 0.2586, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.8850547159398155e-05, | |
| "loss": 0.2407, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.878086307914439e-05, | |
| "loss": 0.2518, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.871117899889063e-05, | |
| "loss": 0.254, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.8641494918636874e-05, | |
| "loss": 0.2517, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.857181083838311e-05, | |
| "loss": 0.248, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.850212675812935e-05, | |
| "loss": 0.2483, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.8432442677875585e-05, | |
| "loss": 0.2465, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.836275859762182e-05, | |
| "loss": 0.243, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.829307451736806e-05, | |
| "loss": 0.2518, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.82233904371143e-05, | |
| "loss": 0.2548, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.815370635686054e-05, | |
| "loss": 0.2492, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.808402227660678e-05, | |
| "loss": 0.2444, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.8014338196353015e-05, | |
| "loss": 0.2436, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.794465411609925e-05, | |
| "loss": 0.2494, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.787497003584549e-05, | |
| "loss": 0.2595, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.7805285955591727e-05, | |
| "loss": 0.2507, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.7735601875337964e-05, | |
| "loss": 0.2388, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.766591779508421e-05, | |
| "loss": 0.2455, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.7596233714830445e-05, | |
| "loss": 0.2364, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.752654963457668e-05, | |
| "loss": 0.2433, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.7456865554322926e-05, | |
| "loss": 0.2519, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.738718147406916e-05, | |
| "loss": 0.2495, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.73174973938154e-05, | |
| "loss": 0.2405, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.7247813313561644e-05, | |
| "loss": 0.2431, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.717812923330788e-05, | |
| "loss": 0.2474, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.710844515305412e-05, | |
| "loss": 0.2468, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.7038761072800356e-05, | |
| "loss": 0.2464, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.696907699254659e-05, | |
| "loss": 0.2464, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.689939291229283e-05, | |
| "loss": 0.2424, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.682970883203907e-05, | |
| "loss": 0.2379, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.676002475178531e-05, | |
| "loss": 0.2442, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.669034067153155e-05, | |
| "loss": 0.246, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.6620656591277786e-05, | |
| "loss": 0.2403, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.655097251102402e-05, | |
| "loss": 0.2429, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.648128843077026e-05, | |
| "loss": 0.2482, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.64116043505165e-05, | |
| "loss": 0.2432, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.6341920270262735e-05, | |
| "loss": 0.2374, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.627223619000898e-05, | |
| "loss": 0.2539, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.6202552109755216e-05, | |
| "loss": 0.2496, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.613286802950145e-05, | |
| "loss": 0.2447, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.606318394924769e-05, | |
| "loss": 0.2396, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.599349986899393e-05, | |
| "loss": 0.2472, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.5923815788740164e-05, | |
| "loss": 0.2431, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.585413170848641e-05, | |
| "loss": 0.2414, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.5784447628232646e-05, | |
| "loss": 0.2398, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.571476354797888e-05, | |
| "loss": 0.2415, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.564507946772512e-05, | |
| "loss": 0.2428, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.5575395387471364e-05, | |
| "loss": 0.2341, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.55057113072176e-05, | |
| "loss": 0.2461, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.543602722696384e-05, | |
| "loss": 0.2481, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.536634314671008e-05, | |
| "loss": 0.2409, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.529665906645632e-05, | |
| "loss": 0.2388, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.522697498620256e-05, | |
| "loss": 0.2427, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.5157290905948794e-05, | |
| "loss": 0.238, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.508760682569503e-05, | |
| "loss": 0.2453, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.501792274544127e-05, | |
| "loss": 0.245, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.4948238665187505e-05, | |
| "loss": 0.2395, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.487855458493375e-05, | |
| "loss": 0.2369, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.4808870504679986e-05, | |
| "loss": 0.2369, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.4739186424426224e-05, | |
| "loss": 0.2476, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.466950234417246e-05, | |
| "loss": 0.238, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.45998182639187e-05, | |
| "loss": 0.2417, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.4530134183664935e-05, | |
| "loss": 0.2433, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.446045010341118e-05, | |
| "loss": 0.2431, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.4390766023157416e-05, | |
| "loss": 0.2434, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.4321081942903654e-05, | |
| "loss": 0.2459, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.425139786264989e-05, | |
| "loss": 0.2464, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.418171378239613e-05, | |
| "loss": 0.2357, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.4112029702142365e-05, | |
| "loss": 0.2394, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.40423456218886e-05, | |
| "loss": 0.2404, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.3972661541634846e-05, | |
| "loss": 0.2333, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.3902977461381083e-05, | |
| "loss": 0.2323, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.383329338112732e-05, | |
| "loss": 0.2562, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.376360930087356e-05, | |
| "loss": 0.238, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.36939252206198e-05, | |
| "loss": 0.2409, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.362424114036604e-05, | |
| "loss": 0.2435, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.3554557060112276e-05, | |
| "loss": 0.2377, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.348487297985852e-05, | |
| "loss": 0.2287, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.341518889960476e-05, | |
| "loss": 0.2427, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.3345504819350994e-05, | |
| "loss": 0.2438, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.327582073909723e-05, | |
| "loss": 0.2447, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.320613665884347e-05, | |
| "loss": 0.2375, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.3136452578589706e-05, | |
| "loss": 0.2399, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.306676849833595e-05, | |
| "loss": 0.2416, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.299708441808219e-05, | |
| "loss": 0.2405, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.2927400337828424e-05, | |
| "loss": 0.2322, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.285771625757466e-05, | |
| "loss": 0.2337, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.27880321773209e-05, | |
| "loss": 0.2391, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.2718348097067136e-05, | |
| "loss": 0.239, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.264866401681337e-05, | |
| "loss": 0.2363, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.257897993655962e-05, | |
| "loss": 0.239, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.2509295856305854e-05, | |
| "loss": 0.243, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.243961177605209e-05, | |
| "loss": 0.2338, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.236992769579833e-05, | |
| "loss": 0.2386, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.2300243615544566e-05, | |
| "loss": 0.2339, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.22305595352908e-05, | |
| "loss": 0.2372, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.216087545503705e-05, | |
| "loss": 0.2368, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.2091191374783284e-05, | |
| "loss": 0.2344, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.202150729452952e-05, | |
| "loss": 0.2402, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.195182321427576e-05, | |
| "loss": 0.2352, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.1882139134021996e-05, | |
| "loss": 0.2358, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.181245505376824e-05, | |
| "loss": 0.2431, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.174277097351448e-05, | |
| "loss": 0.2312, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.1673086893260714e-05, | |
| "loss": 0.2384, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.160340281300696e-05, | |
| "loss": 0.2352, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.1533718732753195e-05, | |
| "loss": 0.236, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.146403465249943e-05, | |
| "loss": 0.2491, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.139435057224567e-05, | |
| "loss": 0.2431, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.132466649199191e-05, | |
| "loss": 0.2351, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.1254982411738144e-05, | |
| "loss": 0.2347, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.118529833148439e-05, | |
| "loss": 0.2349, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.1115614251230625e-05, | |
| "loss": 0.2296, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.104593017097686e-05, | |
| "loss": 0.2408, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.09762460907231e-05, | |
| "loss": 0.2287, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.0906562010469337e-05, | |
| "loss": 0.2327, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.0836877930215574e-05, | |
| "loss": 0.2329, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.076719384996182e-05, | |
| "loss": 0.2401, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.0697509769708055e-05, | |
| "loss": 0.2366, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.062782568945429e-05, | |
| "loss": 0.2351, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.055814160920053e-05, | |
| "loss": 0.236, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.0488457528946766e-05, | |
| "loss": 0.2282, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.0418773448693007e-05, | |
| "loss": 0.2339, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.0349089368439244e-05, | |
| "loss": 0.2336, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.0279405288185485e-05, | |
| "loss": 0.2384, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.0209721207931725e-05, | |
| "loss": 0.2247, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.0140037127677963e-05, | |
| "loss": 0.2345, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.00703530474242e-05, | |
| "loss": 0.2296, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.0000668967170437e-05, | |
| "loss": 0.2319, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.9930984886916674e-05, | |
| "loss": 0.2454, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.986130080666291e-05, | |
| "loss": 0.234, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 2.9791616726409155e-05, | |
| "loss": 0.2305, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.9721932646155392e-05, | |
| "loss": 0.2346, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.965224856590163e-05, | |
| "loss": 0.2336, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.9582564485647867e-05, | |
| "loss": 0.2312, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.9512880405394104e-05, | |
| "loss": 0.2349, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.9443196325140345e-05, | |
| "loss": 0.2343, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.9373512244886585e-05, | |
| "loss": 0.23, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.9303828164632822e-05, | |
| "loss": 0.2221, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.9234144084379063e-05, | |
| "loss": 0.2368, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.91644600041253e-05, | |
| "loss": 0.2262, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.9094775923871537e-05, | |
| "loss": 0.2276, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.9025091843617774e-05, | |
| "loss": 0.2368, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.895540776336401e-05, | |
| "loss": 0.2373, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.8885723683110256e-05, | |
| "loss": 0.2397, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.8816039602856493e-05, | |
| "loss": 0.2328, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.874635552260273e-05, | |
| "loss": 0.2284, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.8676671442348967e-05, | |
| "loss": 0.229, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.8606987362095204e-05, | |
| "loss": 0.228, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.853730328184144e-05, | |
| "loss": 0.2335, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.8467619201587682e-05, | |
| "loss": 0.2281, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.8397935121333923e-05, | |
| "loss": 0.2329, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.8328251041080163e-05, | |
| "loss": 0.2294, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.82585669608264e-05, | |
| "loss": 0.2292, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.8188882880572638e-05, | |
| "loss": 0.23, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.8119198800318875e-05, | |
| "loss": 0.2255, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.8049514720065112e-05, | |
| "loss": 0.235, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.7979830639811356e-05, | |
| "loss": 0.2368, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.7910146559557593e-05, | |
| "loss": 0.2277, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.784046247930383e-05, | |
| "loss": 0.2381, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.7770778399050068e-05, | |
| "loss": 0.2302, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7701094318796305e-05, | |
| "loss": 0.225, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7631410238542542e-05, | |
| "loss": 0.227, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7561726158288782e-05, | |
| "loss": 0.2362, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7492042078035023e-05, | |
| "loss": 0.2283, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.742235799778126e-05, | |
| "loss": 0.2201, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.73526739175275e-05, | |
| "loss": 0.2327, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7282989837273738e-05, | |
| "loss": 0.2236, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.7213305757019975e-05, | |
| "loss": 0.2362, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.7143621676766212e-05, | |
| "loss": 0.2282, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.707393759651245e-05, | |
| "loss": 0.2388, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.7004253516258693e-05, | |
| "loss": 0.2274, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.693456943600493e-05, | |
| "loss": 0.2294, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.6864885355751168e-05, | |
| "loss": 0.2206, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.6795201275497405e-05, | |
| "loss": 0.2253, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.6725517195243642e-05, | |
| "loss": 0.2387, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.665583311498988e-05, | |
| "loss": 0.232, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.6586149034736123e-05, | |
| "loss": 0.2296, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.651646495448236e-05, | |
| "loss": 0.2168, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.6446780874228598e-05, | |
| "loss": 0.2217, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.637709679397484e-05, | |
| "loss": 0.2333, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.6307412713721076e-05, | |
| "loss": 0.2269, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.6237728633467313e-05, | |
| "loss": 0.2326, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.616804455321355e-05, | |
| "loss": 0.2262, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.6098360472959794e-05, | |
| "loss": 0.2245, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.602867639270603e-05, | |
| "loss": 0.2269, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.5958992312452268e-05, | |
| "loss": 0.2313, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.5889308232198505e-05, | |
| "loss": 0.2239, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.5819624151944743e-05, | |
| "loss": 0.2328, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.574994007169098e-05, | |
| "loss": 0.2284, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.568025599143722e-05, | |
| "loss": 0.2314, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.561057191118346e-05, | |
| "loss": 0.2284, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.5540887830929698e-05, | |
| "loss": 0.216, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.547120375067594e-05, | |
| "loss": 0.2232, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.5401519670422176e-05, | |
| "loss": 0.2267, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.5331835590168413e-05, | |
| "loss": 0.227, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.526215150991465e-05, | |
| "loss": 0.2336, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.5192467429660894e-05, | |
| "loss": 0.2279, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.512278334940713e-05, | |
| "loss": 0.2236, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.505309926915337e-05, | |
| "loss": 0.2309, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.4983415188899606e-05, | |
| "loss": 0.2229, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.4913731108645843e-05, | |
| "loss": 0.2181, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.4844047028392083e-05, | |
| "loss": 0.2312, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.477436294813832e-05, | |
| "loss": 0.2264, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.4704678867884558e-05, | |
| "loss": 0.2313, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.46349947876308e-05, | |
| "loss": 0.2267, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.4565310707377036e-05, | |
| "loss": 0.234, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.4495626627123276e-05, | |
| "loss": 0.2303, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.4425942546869513e-05, | |
| "loss": 0.2201, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.4356258466615754e-05, | |
| "loss": 0.23, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.428657438636199e-05, | |
| "loss": 0.2192, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.421689030610823e-05, | |
| "loss": 0.2319, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.414720622585447e-05, | |
| "loss": 0.226, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.4077522145600706e-05, | |
| "loss": 0.232, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.4007838065346943e-05, | |
| "loss": 0.2327, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.3938153985093184e-05, | |
| "loss": 0.2231, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.386846990483942e-05, | |
| "loss": 0.22, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.3798785824585658e-05, | |
| "loss": 0.2243, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.3729101744331895e-05, | |
| "loss": 0.2287, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.3659417664078136e-05, | |
| "loss": 0.2227, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.3589733583824377e-05, | |
| "loss": 0.2321, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.3520049503570614e-05, | |
| "loss": 0.22, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.3450365423316854e-05, | |
| "loss": 0.2247, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.338068134306309e-05, | |
| "loss": 0.2251, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.331099726280933e-05, | |
| "loss": 0.2182, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.324131318255557e-05, | |
| "loss": 0.2244, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.3171629102301806e-05, | |
| "loss": 0.2236, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.3101945022048044e-05, | |
| "loss": 0.2228, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.303226094179428e-05, | |
| "loss": 0.216, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.296257686154052e-05, | |
| "loss": 0.2137, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.289289278128676e-05, | |
| "loss": 0.2207, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.2823208701032996e-05, | |
| "loss": 0.2285, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.2753524620779236e-05, | |
| "loss": 0.2209, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.2683840540525473e-05, | |
| "loss": 0.2313, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.2614156460271714e-05, | |
| "loss": 0.2304, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.2544472380017955e-05, | |
| "loss": 0.2236, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.2474788299764192e-05, | |
| "loss": 0.2211, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.240510421951043e-05, | |
| "loss": 0.2294, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.2335420139256666e-05, | |
| "loss": 0.221, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.2265736059002907e-05, | |
| "loss": 0.2227, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.2196051978749144e-05, | |
| "loss": 0.2204, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.212636789849538e-05, | |
| "loss": 0.2248, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.2056683818241622e-05, | |
| "loss": 0.2206, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.198699973798786e-05, | |
| "loss": 0.2322, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.1917315657734096e-05, | |
| "loss": 0.2211, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.1847631577480337e-05, | |
| "loss": 0.2307, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.1777947497226574e-05, | |
| "loss": 0.2143, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.1708263416972814e-05, | |
| "loss": 0.2201, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.163857933671905e-05, | |
| "loss": 0.2269, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.1568895256465292e-05, | |
| "loss": 0.2222, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.149921117621153e-05, | |
| "loss": 0.2158, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.1429527095957767e-05, | |
| "loss": 0.216, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.1359843015704007e-05, | |
| "loss": 0.2322, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.1290158935450244e-05, | |
| "loss": 0.2167, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.122047485519648e-05, | |
| "loss": 0.2231, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.1150790774942722e-05, | |
| "loss": 0.2242, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.108110669468896e-05, | |
| "loss": 0.237, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.1011422614435196e-05, | |
| "loss": 0.221, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.0941738534181434e-05, | |
| "loss": 0.2251, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.0872054453927674e-05, | |
| "loss": 0.2272, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.080237037367391e-05, | |
| "loss": 0.2235, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0732686293420152e-05, | |
| "loss": 0.2179, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0663002213166392e-05, | |
| "loss": 0.2276, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.059331813291263e-05, | |
| "loss": 0.2215, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0523634052658867e-05, | |
| "loss": 0.2266, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0453949972405107e-05, | |
| "loss": 0.2185, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0384265892151345e-05, | |
| "loss": 0.2255, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0314581811897582e-05, | |
| "loss": 0.2307, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.024489773164382e-05, | |
| "loss": 0.2201, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.017521365139006e-05, | |
| "loss": 0.2208, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0105529571136297e-05, | |
| "loss": 0.2177, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0035845490882534e-05, | |
| "loss": 0.2136, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.9966161410628775e-05, | |
| "loss": 0.2254, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.9896477330375012e-05, | |
| "loss": 0.2208, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.982679325012125e-05, | |
| "loss": 0.2213, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.975710916986749e-05, | |
| "loss": 0.2253, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.968742508961373e-05, | |
| "loss": 0.2172, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9617741009359967e-05, | |
| "loss": 0.2342, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9548056929106204e-05, | |
| "loss": 0.2177, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9478372848852445e-05, | |
| "loss": 0.2263, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9408688768598682e-05, | |
| "loss": 0.2334, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.933900468834492e-05, | |
| "loss": 0.2152, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.926932060809116e-05, | |
| "loss": 0.2154, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9199636527837397e-05, | |
| "loss": 0.2123, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9129952447583634e-05, | |
| "loss": 0.2204, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9060268367329875e-05, | |
| "loss": 0.2167, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8990584287076112e-05, | |
| "loss": 0.2234, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.892090020682235e-05, | |
| "loss": 0.2244, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.885121612656859e-05, | |
| "loss": 0.2227, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8781532046314827e-05, | |
| "loss": 0.2177, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8711847966061068e-05, | |
| "loss": 0.2212, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8642163885807305e-05, | |
| "loss": 0.2218, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8572479805553545e-05, | |
| "loss": 0.2278, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8502795725299782e-05, | |
| "loss": 0.2233, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.843311164504602e-05, | |
| "loss": 0.2242, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.836342756479226e-05, | |
| "loss": 0.2268, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8293743484538497e-05, | |
| "loss": 0.2189, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.8224059404284735e-05, | |
| "loss": 0.2257, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.8154375324030972e-05, | |
| "loss": 0.2134, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.8084691243777212e-05, | |
| "loss": 0.2147, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.801500716352345e-05, | |
| "loss": 0.2204, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7945323083269687e-05, | |
| "loss": 0.215, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7875639003015927e-05, | |
| "loss": 0.2182, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7805954922762168e-05, | |
| "loss": 0.2259, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.7736270842508405e-05, | |
| "loss": 0.2207, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.7666586762254646e-05, | |
| "loss": 0.2197, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.7596902682000883e-05, | |
| "loss": 0.2229, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.752721860174712e-05, | |
| "loss": 0.2196, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.7457534521493357e-05, | |
| "loss": 0.217, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.7387850441239598e-05, | |
| "loss": 0.2143, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.7318166360985835e-05, | |
| "loss": 0.2177, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.7248482280732072e-05, | |
| "loss": 0.2198, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.7178798200478313e-05, | |
| "loss": 0.2264, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.710911412022455e-05, | |
| "loss": 0.2157, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.7039430039970787e-05, | |
| "loss": 0.2275, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.6969745959717028e-05, | |
| "loss": 0.2197, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.6900061879463265e-05, | |
| "loss": 0.2203, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.6830377799209505e-05, | |
| "loss": 0.2242, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.6760693718955743e-05, | |
| "loss": 0.2214, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.6691009638701983e-05, | |
| "loss": 0.2226, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.662132555844822e-05, | |
| "loss": 0.2247, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.6551641478194458e-05, | |
| "loss": 0.2177, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.6481957397940698e-05, | |
| "loss": 0.215, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.6412273317686935e-05, | |
| "loss": 0.211, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.6342589237433172e-05, | |
| "loss": 0.2142, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.6272905157179413e-05, | |
| "loss": 0.2119, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.620322107692565e-05, | |
| "loss": 0.2178, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.6133536996671887e-05, | |
| "loss": 0.2145, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.6063852916418125e-05, | |
| "loss": 0.2108, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5994168836164365e-05, | |
| "loss": 0.2207, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5924484755910606e-05, | |
| "loss": 0.2127, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5854800675656843e-05, | |
| "loss": 0.2249, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5785116595403084e-05, | |
| "loss": 0.2257, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.571543251514932e-05, | |
| "loss": 0.2178, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5645748434895558e-05, | |
| "loss": 0.2237, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.55760643546418e-05, | |
| "loss": 0.2141, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5506380274388036e-05, | |
| "loss": 0.2165, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5436696194134273e-05, | |
| "loss": 0.2096, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5367012113880513e-05, | |
| "loss": 0.2096, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.529732803362675e-05, | |
| "loss": 0.2258, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.522764395337299e-05, | |
| "loss": 0.2141, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5157959873119227e-05, | |
| "loss": 0.2116, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5088275792865467e-05, | |
| "loss": 0.2153, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5018591712611704e-05, | |
| "loss": 0.218, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.4948907632357942e-05, | |
| "loss": 0.2172, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.4879223552104182e-05, | |
| "loss": 0.2281, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.480953947185042e-05, | |
| "loss": 0.2171, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4739855391596658e-05, | |
| "loss": 0.2155, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4670171311342897e-05, | |
| "loss": 0.2148, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4600487231089136e-05, | |
| "loss": 0.2189, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4530803150835373e-05, | |
| "loss": 0.2221, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.446111907058161e-05, | |
| "loss": 0.2223, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4391434990327851e-05, | |
| "loss": 0.2191, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4321750910074088e-05, | |
| "loss": 0.2195, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4252066829820327e-05, | |
| "loss": 0.2182, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4182382749566568e-05, | |
| "loss": 0.2169, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4112698669312805e-05, | |
| "loss": 0.2239, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4043014589059042e-05, | |
| "loss": 0.2209, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.3973330508805283e-05, | |
| "loss": 0.2043, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.390364642855152e-05, | |
| "loss": 0.2211, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.3833962348297757e-05, | |
| "loss": 0.2159, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.3764278268043996e-05, | |
| "loss": 0.2114, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.3694594187790236e-05, | |
| "loss": 0.2229, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.3624910107536474e-05, | |
| "loss": 0.2158, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.355522602728271e-05, | |
| "loss": 0.2145, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.3485541947028951e-05, | |
| "loss": 0.2132, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.3415857866775188e-05, | |
| "loss": 0.2184, | |
| "step": 262500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.3346173786521426e-05, | |
| "loss": 0.2164, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.3276489706267666e-05, | |
| "loss": 0.2111, | |
| "step": 263500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.3206805626013905e-05, | |
| "loss": 0.2128, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.3137121545760142e-05, | |
| "loss": 0.2203, | |
| "step": 264500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.306743746550638e-05, | |
| "loss": 0.2163, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.299775338525262e-05, | |
| "loss": 0.2169, | |
| "step": 265500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.2928069304998857e-05, | |
| "loss": 0.2157, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.2858385224745096e-05, | |
| "loss": 0.2246, | |
| "step": 266500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.2788701144491335e-05, | |
| "loss": 0.2204, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2719017064237574e-05, | |
| "loss": 0.2137, | |
| "step": 267500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2649332983983811e-05, | |
| "loss": 0.2195, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2579648903730052e-05, | |
| "loss": 0.2189, | |
| "step": 268500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2509964823476289e-05, | |
| "loss": 0.208, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2440280743222526e-05, | |
| "loss": 0.2109, | |
| "step": 269500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2370596662968765e-05, | |
| "loss": 0.2249, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2300912582715004e-05, | |
| "loss": 0.2159, | |
| "step": 270500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2231228502461243e-05, | |
| "loss": 0.2114, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2161544422207482e-05, | |
| "loss": 0.2095, | |
| "step": 271500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2091860341953719e-05, | |
| "loss": 0.218, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2022176261699958e-05, | |
| "loss": 0.2208, | |
| "step": 272500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.1952492181446195e-05, | |
| "loss": 0.2098, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.1882808101192434e-05, | |
| "loss": 0.2073, | |
| "step": 273500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.1813124020938674e-05, | |
| "loss": 0.2084, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.1743439940684911e-05, | |
| "loss": 0.2071, | |
| "step": 274500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.167375586043115e-05, | |
| "loss": 0.2207, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.1604071780177387e-05, | |
| "loss": 0.2186, | |
| "step": 275500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.1534387699923626e-05, | |
| "loss": 0.2132, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.1464703619669865e-05, | |
| "loss": 0.2107, | |
| "step": 276500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.1395019539416104e-05, | |
| "loss": 0.2157, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.1325335459162343e-05, | |
| "loss": 0.2138, | |
| "step": 277500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.1255651378908582e-05, | |
| "loss": 0.2157, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.1185967298654819e-05, | |
| "loss": 0.2101, | |
| "step": 278500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.1116283218401058e-05, | |
| "loss": 0.2177, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.1046599138147295e-05, | |
| "loss": 0.2142, | |
| "step": 279500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.0976915057893534e-05, | |
| "loss": 0.2122, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.0907230977639773e-05, | |
| "loss": 0.2212, | |
| "step": 280500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.0837546897386012e-05, | |
| "loss": 0.2096, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.076786281713225e-05, | |
| "loss": 0.2091, | |
| "step": 281500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0698178736878488e-05, | |
| "loss": 0.2124, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0628494656624727e-05, | |
| "loss": 0.2175, | |
| "step": 282500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0558810576370966e-05, | |
| "loss": 0.2124, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0489126496117203e-05, | |
| "loss": 0.2046, | |
| "step": 283500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0419442415863442e-05, | |
| "loss": 0.2072, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.034975833560968e-05, | |
| "loss": 0.2181, | |
| "step": 284500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.028007425535592e-05, | |
| "loss": 0.22, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0210390175102158e-05, | |
| "loss": 0.2087, | |
| "step": 285500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0140706094848395e-05, | |
| "loss": 0.2074, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0071022014594634e-05, | |
| "loss": 0.211, | |
| "step": 286500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0001337934340872e-05, | |
| "loss": 0.2156, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.931653854087112e-06, | |
| "loss": 0.2081, | |
| "step": 287500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.861969773833351e-06, | |
| "loss": 0.2158, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.792285693579588e-06, | |
| "loss": 0.2212, | |
| "step": 288500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.722601613325827e-06, | |
| "loss": 0.2113, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.652917533072064e-06, | |
| "loss": 0.2052, | |
| "step": 289500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.583233452818303e-06, | |
| "loss": 0.2151, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.513549372564542e-06, | |
| "loss": 0.2147, | |
| "step": 290500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.44386529231078e-06, | |
| "loss": 0.2114, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.37418121205702e-06, | |
| "loss": 0.2175, | |
| "step": 291500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.304497131803257e-06, | |
| "loss": 0.2145, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.234813051549496e-06, | |
| "loss": 0.212, | |
| "step": 292500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.165128971295735e-06, | |
| "loss": 0.2117, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.095444891041972e-06, | |
| "loss": 0.2065, | |
| "step": 293500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.02576081078821e-06, | |
| "loss": 0.2192, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.95607673053445e-06, | |
| "loss": 0.2074, | |
| "step": 294500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.886392650280688e-06, | |
| "loss": 0.2155, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.816708570026927e-06, | |
| "loss": 0.2099, | |
| "step": 295500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.747024489773165e-06, | |
| "loss": 0.2192, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.677340409519403e-06, | |
| "loss": 0.2117, | |
| "step": 296500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.60765632926564e-06, | |
| "loss": 0.2118, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.53797224901188e-06, | |
| "loss": 0.2074, | |
| "step": 297500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.468288168758118e-06, | |
| "loss": 0.2071, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.398604088504357e-06, | |
| "loss": 0.2113, | |
| "step": 298500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.328920008250596e-06, | |
| "loss": 0.2072, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.259235927996833e-06, | |
| "loss": 0.2048, | |
| "step": 299500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.189551847743072e-06, | |
| "loss": 0.2104, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.119867767489311e-06, | |
| "loss": 0.218, | |
| "step": 300500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.050183687235548e-06, | |
| "loss": 0.2168, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.980499606981789e-06, | |
| "loss": 0.2217, | |
| "step": 301500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.910815526728026e-06, | |
| "loss": 0.2126, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.841131446474265e-06, | |
| "loss": 0.2084, | |
| "step": 302500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.771447366220504e-06, | |
| "loss": 0.2098, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.701763285966741e-06, | |
| "loss": 0.2131, | |
| "step": 303500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.63207920571298e-06, | |
| "loss": 0.2078, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.562395125459218e-06, | |
| "loss": 0.2164, | |
| "step": 304500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.492711045205457e-06, | |
| "loss": 0.2114, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.423026964951696e-06, | |
| "loss": 0.206, | |
| "step": 305500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.353342884697934e-06, | |
| "loss": 0.2108, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.2836588044441725e-06, | |
| "loss": 0.2156, | |
| "step": 306500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.2139747241904106e-06, | |
| "loss": 0.2124, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.1442906439366494e-06, | |
| "loss": 0.2197, | |
| "step": 307500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.074606563682888e-06, | |
| "loss": 0.2127, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.0049224834291255e-06, | |
| "loss": 0.2113, | |
| "step": 308500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.935238403175364e-06, | |
| "loss": 0.2093, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.865554322921602e-06, | |
| "loss": 0.2191, | |
| "step": 309500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.795870242667841e-06, | |
| "loss": 0.2023, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.72618616241408e-06, | |
| "loss": 0.2092, | |
| "step": 310500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.656502082160318e-06, | |
| "loss": 0.2088, | |
| "step": 311000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.586818001906557e-06, | |
| "loss": 0.2077, | |
| "step": 311500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.517133921652794e-06, | |
| "loss": 0.2075, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.447449841399034e-06, | |
| "loss": 0.2123, | |
| "step": 312500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.377765761145273e-06, | |
| "loss": 0.2137, | |
| "step": 313000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.30808168089151e-06, | |
| "loss": 0.2154, | |
| "step": 313500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.238397600637749e-06, | |
| "loss": 0.2137, | |
| "step": 314000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.168713520383988e-06, | |
| "loss": 0.2106, | |
| "step": 314500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.099029440130226e-06, | |
| "loss": 0.2101, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.029345359876464e-06, | |
| "loss": 0.2154, | |
| "step": 315500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.959661279622703e-06, | |
| "loss": 0.2122, | |
| "step": 316000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.889977199368942e-06, | |
| "loss": 0.2119, | |
| "step": 316500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.82029311911518e-06, | |
| "loss": 0.2069, | |
| "step": 317000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.750609038861418e-06, | |
| "loss": 0.2166, | |
| "step": 317500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.6809249586076566e-06, | |
| "loss": 0.2135, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.611240878353895e-06, | |
| "loss": 0.216, | |
| "step": 318500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.5415567981001335e-06, | |
| "loss": 0.2078, | |
| "step": 319000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.471872717846372e-06, | |
| "loss": 0.2033, | |
| "step": 319500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.40218863759261e-06, | |
| "loss": 0.218, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.3325045573388484e-06, | |
| "loss": 0.2128, | |
| "step": 320500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.262820477085087e-06, | |
| "loss": 0.2115, | |
| "step": 321000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.193136396831326e-06, | |
| "loss": 0.2094, | |
| "step": 321500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.123452316577564e-06, | |
| "loss": 0.2096, | |
| "step": 322000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.053768236323802e-06, | |
| "loss": 0.2107, | |
| "step": 322500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.984084156070041e-06, | |
| "loss": 0.2123, | |
| "step": 323000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.91440007581628e-06, | |
| "loss": 0.2089, | |
| "step": 323500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.844715995562518e-06, | |
| "loss": 0.2103, | |
| "step": 324000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.775031915308757e-06, | |
| "loss": 0.2089, | |
| "step": 324500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.705347835054995e-06, | |
| "loss": 0.2041, | |
| "step": 325000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.635663754801233e-06, | |
| "loss": 0.2033, | |
| "step": 325500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.565979674547472e-06, | |
| "loss": 0.2161, | |
| "step": 326000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.496295594293711e-06, | |
| "loss": 0.2082, | |
| "step": 326500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.426611514039949e-06, | |
| "loss": 0.2064, | |
| "step": 327000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.356927433786187e-06, | |
| "loss": 0.2146, | |
| "step": 327500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.287243353532426e-06, | |
| "loss": 0.2149, | |
| "step": 328000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.2175592732786646e-06, | |
| "loss": 0.2031, | |
| "step": 328500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.147875193024903e-06, | |
| "loss": 0.2051, | |
| "step": 329000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.078191112771141e-06, | |
| "loss": 0.2138, | |
| "step": 329500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.0085070325173795e-06, | |
| "loss": 0.2039, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.9388229522636175e-06, | |
| "loss": 0.2155, | |
| "step": 330500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.869138872009856e-06, | |
| "loss": 0.2039, | |
| "step": 331000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.799454791756095e-06, | |
| "loss": 0.2121, | |
| "step": 331500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.7297707115023333e-06, | |
| "loss": 0.2129, | |
| "step": 332000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.6600866312485714e-06, | |
| "loss": 0.2153, | |
| "step": 332500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.59040255099481e-06, | |
| "loss": 0.2072, | |
| "step": 333000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.5207184707410487e-06, | |
| "loss": 0.2059, | |
| "step": 333500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.451034390487287e-06, | |
| "loss": 0.2111, | |
| "step": 334000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.3813503102335256e-06, | |
| "loss": 0.2161, | |
| "step": 334500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.3116662299797636e-06, | |
| "loss": 0.2153, | |
| "step": 335000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.241982149726002e-06, | |
| "loss": 0.2092, | |
| "step": 335500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.172298069472241e-06, | |
| "loss": 0.2151, | |
| "step": 336000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.1026139892184794e-06, | |
| "loss": 0.2036, | |
| "step": 336500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.032929908964718e-06, | |
| "loss": 0.214, | |
| "step": 337000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.963245828710956e-06, | |
| "loss": 0.214, | |
| "step": 337500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.8935617484571948e-06, | |
| "loss": 0.2063, | |
| "step": 338000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.823877668203433e-06, | |
| "loss": 0.2064, | |
| "step": 338500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.7541935879496717e-06, | |
| "loss": 0.2051, | |
| "step": 339000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.68450950769591e-06, | |
| "loss": 0.2069, | |
| "step": 339500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.614825427442148e-06, | |
| "loss": 0.2058, | |
| "step": 340000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.545141347188387e-06, | |
| "loss": 0.2122, | |
| "step": 340500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.475457266934625e-06, | |
| "loss": 0.202, | |
| "step": 341000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.405773186680864e-06, | |
| "loss": 0.2069, | |
| "step": 341500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.336089106427102e-06, | |
| "loss": 0.2081, | |
| "step": 342000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.2664050261733405e-06, | |
| "loss": 0.2072, | |
| "step": 342500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.1967209459195793e-06, | |
| "loss": 0.2092, | |
| "step": 343000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.1270368656658174e-06, | |
| "loss": 0.2121, | |
| "step": 343500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.0573527854120563e-06, | |
| "loss": 0.2068, | |
| "step": 344000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.9876687051582943e-06, | |
| "loss": 0.201, | |
| "step": 344500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.9179846249045327e-06, | |
| "loss": 0.2171, | |
| "step": 345000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.8483005446507714e-06, | |
| "loss": 0.2043, | |
| "step": 345500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.7786164643970096e-06, | |
| "loss": 0.2038, | |
| "step": 346000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.7089323841432483e-06, | |
| "loss": 0.2107, | |
| "step": 346500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.6392483038894866e-06, | |
| "loss": 0.2078, | |
| "step": 347000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.5695642236357252e-06, | |
| "loss": 0.209, | |
| "step": 347500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.4998801433819637e-06, | |
| "loss": 0.2118, | |
| "step": 348000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.4301960631282021e-06, | |
| "loss": 0.2247, | |
| "step": 348500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.3605119828744404e-06, | |
| "loss": 0.2094, | |
| "step": 349000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.2908279026206788e-06, | |
| "loss": 0.2154, | |
| "step": 349500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.2211438223669173e-06, | |
| "loss": 0.2062, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.151459742113156e-06, | |
| "loss": 0.2105, | |
| "step": 350500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0817756618593944e-06, | |
| "loss": 0.2177, | |
| "step": 351000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.0120915816056329e-06, | |
| "loss": 0.2062, | |
| "step": 351500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.424075013518711e-07, | |
| "loss": 0.205, | |
| "step": 352000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.727234210981097e-07, | |
| "loss": 0.2032, | |
| "step": 352500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.030393408443481e-07, | |
| "loss": 0.2169, | |
| "step": 353000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.333552605905866e-07, | |
| "loss": 0.2153, | |
| "step": 353500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.636711803368249e-07, | |
| "loss": 0.2046, | |
| "step": 354000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.939871000830635e-07, | |
| "loss": 0.2068, | |
| "step": 354500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.243030198293018e-07, | |
| "loss": 0.2188, | |
| "step": 355000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.5461893957554035e-07, | |
| "loss": 0.2139, | |
| "step": 355500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.849348593217788e-07, | |
| "loss": 0.2085, | |
| "step": 356000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.1525077906801726e-07, | |
| "loss": 0.1994, | |
| "step": 356500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.455666988142557e-07, | |
| "loss": 0.216, | |
| "step": 357000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.7588261856049415e-07, | |
| "loss": 0.2123, | |
| "step": 357500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.061985383067326e-07, | |
| "loss": 0.2063, | |
| "step": 358000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.651445805297105e-08, | |
| "loss": 0.2098, | |
| "step": 358500 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 358762, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 1.26740447502336e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |