| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 6717, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.94789081885856e-08, | |
| "loss": 4.9627, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4888337468982628e-07, | |
| "loss": 4.7244, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.28287841191067e-07, | |
| "loss": 4.2955, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.076923076923077e-07, | |
| "loss": 3.4769, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.8709677419354837e-07, | |
| "loss": 2.602, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.665012406947891e-07, | |
| "loss": 1.7352, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.459057071960298e-07, | |
| "loss": 1.5972, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.253101736972705e-07, | |
| "loss": 1.4621, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.047146401985111e-07, | |
| "loss": 1.2769, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 7.841191066997518e-07, | |
| "loss": 0.9435, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 8.635235732009926e-07, | |
| "loss": 0.4583, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.429280397022333e-07, | |
| "loss": 0.4589, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.985745961355717e-07, | |
| "loss": 0.456, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.935064935064936e-07, | |
| "loss": 0.3103, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.884383908774152e-07, | |
| "loss": 0.4253, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.83370288248337e-07, | |
| "loss": 0.363, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.783021856192588e-07, | |
| "loss": 0.2691, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.732340829901805e-07, | |
| "loss": 0.2752, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.681659803611024e-07, | |
| "loss": 0.292, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.63097877732024e-07, | |
| "loss": 0.3344, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.580297751029457e-07, | |
| "loss": 0.3355, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.529616724738676e-07, | |
| "loss": 0.3746, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.478935698447893e-07, | |
| "loss": 0.3536, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.428254672157111e-07, | |
| "loss": 0.2955, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.377573645866328e-07, | |
| "loss": 0.3463, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.326892619575546e-07, | |
| "loss": 0.4049, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.276211593284764e-07, | |
| "loss": 0.3387, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.225530566993981e-07, | |
| "loss": 0.3181, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.174849540703199e-07, | |
| "loss": 0.2915, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.125752296484004e-07, | |
| "loss": 0.298, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.075071270193221e-07, | |
| "loss": 0.3243, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.024390243902439e-07, | |
| "loss": 0.2464, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.973709217611656e-07, | |
| "loss": 0.3151, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.923028191320873e-07, | |
| "loss": 0.3568, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.872347165030091e-07, | |
| "loss": 0.2174, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.821666138739309e-07, | |
| "loss": 0.3157, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.770985112448527e-07, | |
| "loss": 0.3387, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.720304086157745e-07, | |
| "loss": 0.3334, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.669623059866962e-07, | |
| "loss": 0.2684, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.618942033576179e-07, | |
| "loss": 0.2936, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.568261007285396e-07, | |
| "loss": 0.3284, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.517579980994615e-07, | |
| "loss": 0.1762, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.466898954703833e-07, | |
| "loss": 0.3034, | |
| "step": 1376 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.41621792841305e-07, | |
| "loss": 0.3374, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.365536902122268e-07, | |
| "loss": 0.2804, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.314855875831485e-07, | |
| "loss": 0.2602, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.264174849540702e-07, | |
| "loss": 0.3025, | |
| "step": 1504 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.213493823249921e-07, | |
| "loss": 0.3233, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.162812796959138e-07, | |
| "loss": 0.3104, | |
| "step": 1568 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.112131770668356e-07, | |
| "loss": 0.3905, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.061450744377573e-07, | |
| "loss": 0.2828, | |
| "step": 1632 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.01076971808679e-07, | |
| "loss": 0.2781, | |
| "step": 1664 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.960088691796008e-07, | |
| "loss": 0.2269, | |
| "step": 1696 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.909407665505227e-07, | |
| "loss": 0.2633, | |
| "step": 1728 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.858726639214444e-07, | |
| "loss": 0.3526, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.808045612923662e-07, | |
| "loss": 0.2754, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.757364586632878e-07, | |
| "loss": 0.268, | |
| "step": 1824 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.706683560342096e-07, | |
| "loss": 0.3112, | |
| "step": 1856 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.656002534051315e-07, | |
| "loss": 0.2976, | |
| "step": 1888 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.605321507760532e-07, | |
| "loss": 0.2428, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.55464048146975e-07, | |
| "loss": 0.2446, | |
| "step": 1952 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.503959455178968e-07, | |
| "loss": 0.2443, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.453278428888184e-07, | |
| "loss": 0.2407, | |
| "step": 2016 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.402597402597402e-07, | |
| "loss": 0.2782, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.351916376306619e-07, | |
| "loss": 0.5238, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.301235350015838e-07, | |
| "loss": 0.2876, | |
| "step": 2112 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.250554323725056e-07, | |
| "loss": 0.3338, | |
| "step": 2144 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.199873297434272e-07, | |
| "loss": 0.2832, | |
| "step": 2176 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.14919227114349e-07, | |
| "loss": 0.3335, | |
| "step": 2208 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.098511244852708e-07, | |
| "loss": 0.2526, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.047830218561926e-07, | |
| "loss": 0.2588, | |
| "step": 2272 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.99873297434273e-07, | |
| "loss": 0.3761, | |
| "step": 2304 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.948051948051947e-07, | |
| "loss": 0.2686, | |
| "step": 2336 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.897370921761165e-07, | |
| "loss": 0.2223, | |
| "step": 2368 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.846689895470384e-07, | |
| "loss": 0.3035, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.796008869179601e-07, | |
| "loss": 0.3097, | |
| "step": 2432 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.745327842888818e-07, | |
| "loss": 0.2795, | |
| "step": 2464 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.694646816598036e-07, | |
| "loss": 0.2274, | |
| "step": 2496 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.643965790307253e-07, | |
| "loss": 0.2918, | |
| "step": 2528 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.593284764016471e-07, | |
| "loss": 0.2982, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.542603737725689e-07, | |
| "loss": 0.2517, | |
| "step": 2592 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.491922711434906e-07, | |
| "loss": 0.2184, | |
| "step": 2624 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.441241685144124e-07, | |
| "loss": 0.2715, | |
| "step": 2656 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.390560658853341e-07, | |
| "loss": 0.2155, | |
| "step": 2688 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.339879632562559e-07, | |
| "loss": 0.2686, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.289198606271777e-07, | |
| "loss": 0.2818, | |
| "step": 2752 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.238517579980995e-07, | |
| "loss": 0.2865, | |
| "step": 2784 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.187836553690212e-07, | |
| "loss": 0.2123, | |
| "step": 2816 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.137155527399429e-07, | |
| "loss": 0.2663, | |
| "step": 2848 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.086474501108647e-07, | |
| "loss": 0.2821, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.035793474817865e-07, | |
| "loss": 0.3285, | |
| "step": 2912 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.985112448527082e-07, | |
| "loss": 0.2123, | |
| "step": 2944 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.934431422236301e-07, | |
| "loss": 0.2753, | |
| "step": 2976 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.883750395945518e-07, | |
| "loss": 0.2082, | |
| "step": 3008 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.833069369654735e-07, | |
| "loss": 0.2989, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.782388343363953e-07, | |
| "loss": 0.2268, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.73170731707317e-07, | |
| "loss": 0.2775, | |
| "step": 3104 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.681026290782388e-07, | |
| "loss": 0.2643, | |
| "step": 3136 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.630345264491606e-07, | |
| "loss": 0.1845, | |
| "step": 3168 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.579664238200823e-07, | |
| "loss": 0.282, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.528983211910041e-07, | |
| "loss": 0.3109, | |
| "step": 3232 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.478302185619259e-07, | |
| "loss": 0.2203, | |
| "step": 3264 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.427621159328476e-07, | |
| "loss": 0.2803, | |
| "step": 3296 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.376940133037694e-07, | |
| "loss": 0.298, | |
| "step": 3328 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.326259106746911e-07, | |
| "loss": 0.2953, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.275578080456129e-07, | |
| "loss": 0.2844, | |
| "step": 3392 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.224897054165347e-07, | |
| "loss": 0.2202, | |
| "step": 3424 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.174216027874564e-07, | |
| "loss": 0.2211, | |
| "step": 3456 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.123535001583782e-07, | |
| "loss": 0.3406, | |
| "step": 3488 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.072853975292999e-07, | |
| "loss": 0.2894, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.022172949002217e-07, | |
| "loss": 0.2595, | |
| "step": 3552 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.971491922711435e-07, | |
| "loss": 0.2943, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.920810896420652e-07, | |
| "loss": 0.2118, | |
| "step": 3616 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.87012987012987e-07, | |
| "loss": 0.2267, | |
| "step": 3648 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.819448843839088e-07, | |
| "loss": 0.2574, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.768767817548305e-07, | |
| "loss": 0.2322, | |
| "step": 3712 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.7180867912575227e-07, | |
| "loss": 0.2427, | |
| "step": 3744 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.6674057649667405e-07, | |
| "loss": 0.252, | |
| "step": 3776 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.616724738675958e-07, | |
| "loss": 0.2369, | |
| "step": 3808 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.5660437123851757e-07, | |
| "loss": 0.2179, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.5153626860943935e-07, | |
| "loss": 0.2108, | |
| "step": 3872 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.464681659803611e-07, | |
| "loss": 0.1787, | |
| "step": 3904 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.414000633512828e-07, | |
| "loss": 0.281, | |
| "step": 3936 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.3633196072220465e-07, | |
| "loss": 0.3145, | |
| "step": 3968 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.312638580931264e-07, | |
| "loss": 0.2586, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.261957554640481e-07, | |
| "loss": 0.2548, | |
| "step": 4032 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.211276528349699e-07, | |
| "loss": 0.3165, | |
| "step": 4064 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.160595502058917e-07, | |
| "loss": 0.1822, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.109914475768134e-07, | |
| "loss": 0.1949, | |
| "step": 4128 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.059233449477352e-07, | |
| "loss": 0.3197, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.008552423186569e-07, | |
| "loss": 0.2209, | |
| "step": 4192 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.957871396895787e-07, | |
| "loss": 0.1941, | |
| "step": 4224 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.907190370605005e-07, | |
| "loss": 0.2412, | |
| "step": 4256 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.856509344314222e-07, | |
| "loss": 0.3301, | |
| "step": 4288 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.8058283180234395e-07, | |
| "loss": 0.3845, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.755147291732658e-07, | |
| "loss": 0.2557, | |
| "step": 4352 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.704466265441875e-07, | |
| "loss": 0.2228, | |
| "step": 4384 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.6537852391510925e-07, | |
| "loss": 0.2608, | |
| "step": 4416 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.6031042128603103e-07, | |
| "loss": 0.1698, | |
| "step": 4448 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.552423186569528e-07, | |
| "loss": 0.2359, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.5017421602787454e-07, | |
| "loss": 0.2657, | |
| "step": 4512 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.451061133987963e-07, | |
| "loss": 0.2521, | |
| "step": 4544 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.4003801076971806e-07, | |
| "loss": 0.216, | |
| "step": 4576 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.3496990814063984e-07, | |
| "loss": 0.3104, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.299018055115616e-07, | |
| "loss": 0.2136, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.2483370288248335e-07, | |
| "loss": 0.2443, | |
| "step": 4672 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.197656002534051e-07, | |
| "loss": 0.1771, | |
| "step": 4704 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.146974976243269e-07, | |
| "loss": 0.187, | |
| "step": 4736 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0962939499524865e-07, | |
| "loss": 0.2901, | |
| "step": 4768 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.045612923661704e-07, | |
| "loss": 0.4018, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9949318973709216e-07, | |
| "loss": 0.2795, | |
| "step": 4832 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9442508710801395e-07, | |
| "loss": 0.2578, | |
| "step": 4864 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.893569844789357e-07, | |
| "loss": 0.1746, | |
| "step": 4896 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.8428888184985746e-07, | |
| "loss": 0.2534, | |
| "step": 4928 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.792207792207792e-07, | |
| "loss": 0.3677, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.74152676591701e-07, | |
| "loss": 0.2953, | |
| "step": 4992 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6908457396262276e-07, | |
| "loss": 0.2722, | |
| "step": 5024 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.640164713335445e-07, | |
| "loss": 0.2797, | |
| "step": 5056 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.589483687044662e-07, | |
| "loss": 0.2707, | |
| "step": 5088 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5388026607538806e-07, | |
| "loss": 0.3243, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.488121634463098e-07, | |
| "loss": 0.271, | |
| "step": 5152 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.437440608172315e-07, | |
| "loss": 0.2899, | |
| "step": 5184 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.386759581881533e-07, | |
| "loss": 0.3117, | |
| "step": 5216 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3360785555907506e-07, | |
| "loss": 0.2323, | |
| "step": 5248 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2853975292999684e-07, | |
| "loss": 0.2318, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2347165030091857e-07, | |
| "loss": 0.29, | |
| "step": 5312 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1840354767184035e-07, | |
| "loss": 0.2406, | |
| "step": 5344 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.133354450427621e-07, | |
| "loss": 0.2292, | |
| "step": 5376 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0826734241368387e-07, | |
| "loss": 0.3598, | |
| "step": 5408 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0319923978460563e-07, | |
| "loss": 0.258, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.981311371555274e-07, | |
| "loss": 0.2904, | |
| "step": 5472 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.9306303452644914e-07, | |
| "loss": 0.3177, | |
| "step": 5504 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8799493189737092e-07, | |
| "loss": 0.3294, | |
| "step": 5536 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8292682926829268e-07, | |
| "loss": 0.2415, | |
| "step": 5568 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7785872663921444e-07, | |
| "loss": 0.2808, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.727906240101362e-07, | |
| "loss": 0.2563, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6772252138105798e-07, | |
| "loss": 0.2069, | |
| "step": 5664 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.626544187519797e-07, | |
| "loss": 0.2112, | |
| "step": 5696 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.575863161229015e-07, | |
| "loss": 0.1768, | |
| "step": 5728 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.5251821349382325e-07, | |
| "loss": 0.2977, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.47450110864745e-07, | |
| "loss": 0.2511, | |
| "step": 5792 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.4238200823566676e-07, | |
| "loss": 0.2203, | |
| "step": 5824 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.3731390560658854e-07, | |
| "loss": 0.1882, | |
| "step": 5856 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.3224580297751027e-07, | |
| "loss": 0.3103, | |
| "step": 5888 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2717770034843206e-07, | |
| "loss": 0.2475, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2210959771935381e-07, | |
| "loss": 0.2512, | |
| "step": 5952 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1704149509027557e-07, | |
| "loss": 0.2936, | |
| "step": 5984 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1197339246119733e-07, | |
| "loss": 0.2215, | |
| "step": 6016 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0690528983211909e-07, | |
| "loss": 0.2263, | |
| "step": 6048 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0183718720304086e-07, | |
| "loss": 0.2768, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.676908457396261e-08, | |
| "loss": 0.2535, | |
| "step": 6112 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.170098194488437e-08, | |
| "loss": 0.195, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.663287931580614e-08, | |
| "loss": 0.1955, | |
| "step": 6176 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.15647766867279e-08, | |
| "loss": 0.355, | |
| "step": 6208 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.649667405764967e-08, | |
| "loss": 0.2593, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.142857142857142e-08, | |
| "loss": 0.3068, | |
| "step": 6272 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.636046879949318e-08, | |
| "loss": 0.2366, | |
| "step": 6304 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.129236617041495e-08, | |
| "loss": 0.2361, | |
| "step": 6336 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.6224263541336714e-08, | |
| "loss": 0.2616, | |
| "step": 6368 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.115616091225847e-08, | |
| "loss": 0.2597, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.6088058283180234e-08, | |
| "loss": 0.3039, | |
| "step": 6432 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.1019955654102e-08, | |
| "loss": 0.2453, | |
| "step": 6464 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.595185302502376e-08, | |
| "loss": 0.3413, | |
| "step": 6496 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.088375039594552e-08, | |
| "loss": 0.2654, | |
| "step": 6528 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.581564776686728e-08, | |
| "loss": 0.2189, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.074754513778904e-08, | |
| "loss": 0.257, | |
| "step": 6592 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.56794425087108e-08, | |
| "loss": 0.25, | |
| "step": 6624 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.0611339879632562e-08, | |
| "loss": 0.3328, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.543237250554324e-09, | |
| "loss": 0.2794, | |
| "step": 6688 | |
| } | |
| ], | |
| "logging_steps": 32, | |
| "max_steps": 6717, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 6717, | |
| "total_flos": 2.85226794934272e+19, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |