{ "best_metric": 0.027821063995361328, "best_model_checkpoint": "/temp/t5_base-qg-aap-test/checkpoint-100", "epoch": 100.0, "eval_steps": 500, "global_step": 100, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 1.0, "eval_loss": 8.543387413024902, "eval_runtime": 0.22, "eval_samples_per_second": 36.371, "eval_steps_per_second": 4.546, "step": 1 }, { "epoch": 2.0, "eval_loss": 7.301304340362549, "eval_runtime": 0.2288, "eval_samples_per_second": 34.962, "eval_steps_per_second": 4.37, "step": 2 }, { "epoch": 3.0, "eval_loss": 6.199253559112549, "eval_runtime": 0.2323, "eval_samples_per_second": 34.438, "eval_steps_per_second": 4.305, "step": 3 }, { "epoch": 4.0, "eval_loss": 5.28981351852417, "eval_runtime": 0.2358, "eval_samples_per_second": 33.933, "eval_steps_per_second": 4.242, "step": 4 }, { "epoch": 5.0, "eval_loss": 4.522605895996094, "eval_runtime": 0.2354, "eval_samples_per_second": 33.99, "eval_steps_per_second": 4.249, "step": 5 }, { "epoch": 6.0, "eval_loss": 3.9202158451080322, "eval_runtime": 0.2368, "eval_samples_per_second": 33.78, "eval_steps_per_second": 4.222, "step": 6 }, { "epoch": 7.0, "eval_loss": 3.4436142444610596, "eval_runtime": 0.2346, "eval_samples_per_second": 34.095, "eval_steps_per_second": 4.262, "step": 7 }, { "epoch": 8.0, "eval_loss": 3.0407516956329346, "eval_runtime": 0.2359, "eval_samples_per_second": 33.92, "eval_steps_per_second": 4.24, "step": 8 }, { "epoch": 9.0, "eval_loss": 2.7138168811798096, "eval_runtime": 0.232, "eval_samples_per_second": 34.479, "eval_steps_per_second": 4.31, "step": 9 }, { "epoch": 10.0, "eval_loss": 2.443582773208618, "eval_runtime": 0.2358, "eval_samples_per_second": 33.927, "eval_steps_per_second": 4.241, "step": 10 }, { "epoch": 11.0, "eval_loss": 2.2130391597747803, "eval_runtime": 0.2341, "eval_samples_per_second": 34.169, "eval_steps_per_second": 4.271, "step": 11 }, { "epoch": 12.0, "eval_loss": 2.019005060195923, "eval_runtime": 0.2346, "eval_samples_per_second": 34.099, "eval_steps_per_second": 4.262, "step": 12 }, { "epoch": 13.0, "eval_loss": 1.8450862169265747, "eval_runtime": 0.2278, "eval_samples_per_second": 35.113, "eval_steps_per_second": 4.389, "step": 13 }, { "epoch": 14.0, "eval_loss": 1.6746028661727905, "eval_runtime": 0.2332, "eval_samples_per_second": 34.306, "eval_steps_per_second": 4.288, "step": 14 }, { "epoch": 15.0, "eval_loss": 1.5047035217285156, "eval_runtime": 0.2344, "eval_samples_per_second": 34.13, "eval_steps_per_second": 4.266, "step": 15 }, { "epoch": 16.0, "eval_loss": 1.3376126289367676, "eval_runtime": 0.235, "eval_samples_per_second": 34.041, "eval_steps_per_second": 4.255, "step": 16 }, { "epoch": 17.0, "eval_loss": 1.1800236701965332, "eval_runtime": 0.234, "eval_samples_per_second": 34.182, "eval_steps_per_second": 4.273, "step": 17 }, { "epoch": 18.0, "eval_loss": 1.043411374092102, "eval_runtime": 0.2333, "eval_samples_per_second": 34.295, "eval_steps_per_second": 4.287, "step": 18 }, { "epoch": 19.0, "eval_loss": 0.9442155361175537, "eval_runtime": 0.2311, "eval_samples_per_second": 34.61, "eval_steps_per_second": 4.326, "step": 19 }, { "epoch": 20.0, "eval_loss": 0.8739322423934937, "eval_runtime": 0.2326, "eval_samples_per_second": 34.388, "eval_steps_per_second": 4.299, "step": 20 }, { "epoch": 21.0, "eval_loss": 0.8162761926651001, "eval_runtime": 0.2315, "eval_samples_per_second": 34.559, "eval_steps_per_second": 4.32, "step": 21 }, { "epoch": 22.0, "eval_loss": 0.7628746628761292, "eval_runtime": 0.2355, "eval_samples_per_second": 33.977, "eval_steps_per_second": 4.247, "step": 22 }, { "epoch": 23.0, "eval_loss": 0.7117814421653748, "eval_runtime": 0.2347, "eval_samples_per_second": 34.092, "eval_steps_per_second": 4.262, "step": 23 }, { "epoch": 24.0, "eval_loss": 0.661788821220398, "eval_runtime": 0.2304, "eval_samples_per_second": 34.715, "eval_steps_per_second": 4.339, "step": 24 }, { "epoch": 25.0, "eval_loss": 0.6104191541671753, "eval_runtime": 0.2332, "eval_samples_per_second": 34.307, "eval_steps_per_second": 4.288, "step": 25 }, { "epoch": 26.0, "eval_loss": 0.5596593618392944, "eval_runtime": 0.2354, "eval_samples_per_second": 33.987, "eval_steps_per_second": 4.248, "step": 26 }, { "epoch": 27.0, "eval_loss": 0.5112428665161133, "eval_runtime": 0.2316, "eval_samples_per_second": 34.539, "eval_steps_per_second": 4.317, "step": 27 }, { "epoch": 28.0, "eval_loss": 0.4656863212585449, "eval_runtime": 0.2358, "eval_samples_per_second": 33.925, "eval_steps_per_second": 4.241, "step": 28 }, { "epoch": 29.0, "eval_loss": 0.4242892861366272, "eval_runtime": 0.2324, "eval_samples_per_second": 34.417, "eval_steps_per_second": 4.302, "step": 29 }, { "epoch": 30.0, "eval_loss": 0.38730064034461975, "eval_runtime": 0.2361, "eval_samples_per_second": 33.884, "eval_steps_per_second": 4.236, "step": 30 }, { "epoch": 31.0, "eval_loss": 0.3529461622238159, "eval_runtime": 0.2273, "eval_samples_per_second": 35.195, "eval_steps_per_second": 4.399, "step": 31 }, { "epoch": 32.0, "eval_loss": 0.3208673894405365, "eval_runtime": 0.2312, "eval_samples_per_second": 34.6, "eval_steps_per_second": 4.325, "step": 32 }, { "epoch": 33.0, "eval_loss": 0.2917502224445343, "eval_runtime": 0.2299, "eval_samples_per_second": 34.802, "eval_steps_per_second": 4.35, "step": 33 }, { "epoch": 34.0, "eval_loss": 0.26672637462615967, "eval_runtime": 0.2245, "eval_samples_per_second": 35.641, "eval_steps_per_second": 4.455, "step": 34 }, { "epoch": 35.0, "eval_loss": 0.24361155927181244, "eval_runtime": 0.2331, "eval_samples_per_second": 34.323, "eval_steps_per_second": 4.29, "step": 35 }, { "epoch": 36.0, "eval_loss": 0.22150522470474243, "eval_runtime": 0.2314, "eval_samples_per_second": 34.566, "eval_steps_per_second": 4.321, "step": 36 }, { "epoch": 37.0, "eval_loss": 0.20035764575004578, "eval_runtime": 0.2358, "eval_samples_per_second": 33.929, "eval_steps_per_second": 4.241, "step": 37 }, { "epoch": 38.0, "eval_loss": 0.18083304166793823, "eval_runtime": 0.2305, "eval_samples_per_second": 34.715, "eval_steps_per_second": 4.339, "step": 38 }, { "epoch": 39.0, "eval_loss": 0.16373088955879211, "eval_runtime": 0.2358, "eval_samples_per_second": 33.933, "eval_steps_per_second": 4.242, "step": 39 }, { "epoch": 40.0, "eval_loss": 0.14839813113212585, "eval_runtime": 0.236, "eval_samples_per_second": 33.892, "eval_steps_per_second": 4.236, "step": 40 }, { "epoch": 41.0, "eval_loss": 0.1357194185256958, "eval_runtime": 0.2358, "eval_samples_per_second": 33.932, "eval_steps_per_second": 4.242, "step": 41 }, { "epoch": 42.0, "eval_loss": 0.1251770257949829, "eval_runtime": 0.2246, "eval_samples_per_second": 35.621, "eval_steps_per_second": 4.453, "step": 42 }, { "epoch": 43.0, "eval_loss": 0.11590107530355453, "eval_runtime": 0.2355, "eval_samples_per_second": 33.971, "eval_steps_per_second": 4.246, "step": 43 }, { "epoch": 44.0, "eval_loss": 0.10785257816314697, "eval_runtime": 0.2288, "eval_samples_per_second": 34.96, "eval_steps_per_second": 4.37, "step": 44 }, { "epoch": 45.0, "eval_loss": 0.09967806190252304, "eval_runtime": 0.2309, "eval_samples_per_second": 34.652, "eval_steps_per_second": 4.332, "step": 45 }, { "epoch": 46.0, "eval_loss": 0.0922195091843605, "eval_runtime": 0.2356, "eval_samples_per_second": 33.952, "eval_steps_per_second": 4.244, "step": 46 }, { "epoch": 47.0, "eval_loss": 0.08580458164215088, "eval_runtime": 0.2356, "eval_samples_per_second": 33.959, "eval_steps_per_second": 4.245, "step": 47 }, { "epoch": 48.0, "eval_loss": 0.08024830371141434, "eval_runtime": 0.2344, "eval_samples_per_second": 34.129, "eval_steps_per_second": 4.266, "step": 48 }, { "epoch": 49.0, "eval_loss": 0.07487977296113968, "eval_runtime": 0.2357, "eval_samples_per_second": 33.936, "eval_steps_per_second": 4.242, "step": 49 }, { "epoch": 50.0, "eval_loss": 0.07032416760921478, "eval_runtime": 0.231, "eval_samples_per_second": 34.629, "eval_steps_per_second": 4.329, "step": 50 }, { "epoch": 51.0, "eval_loss": 0.06604083627462387, "eval_runtime": 0.2355, "eval_samples_per_second": 33.971, "eval_steps_per_second": 4.246, "step": 51 }, { "epoch": 52.0, "eval_loss": 0.06256596744060516, "eval_runtime": 0.2338, "eval_samples_per_second": 34.216, "eval_steps_per_second": 4.277, "step": 52 }, { "epoch": 53.0, "eval_loss": 0.0596306137740612, "eval_runtime": 0.2355, "eval_samples_per_second": 33.976, "eval_steps_per_second": 4.247, "step": 53 }, { "epoch": 54.0, "eval_loss": 0.05728648975491524, "eval_runtime": 0.2313, "eval_samples_per_second": 34.582, "eval_steps_per_second": 4.323, "step": 54 }, { "epoch": 55.0, "eval_loss": 0.05549569055438042, "eval_runtime": 0.2351, "eval_samples_per_second": 34.022, "eval_steps_per_second": 4.253, "step": 55 }, { "epoch": 56.0, "eval_loss": 0.0538727343082428, "eval_runtime": 0.2351, "eval_samples_per_second": 34.035, "eval_steps_per_second": 4.254, "step": 56 }, { "epoch": 57.0, "eval_loss": 0.05207115411758423, "eval_runtime": 0.2247, "eval_samples_per_second": 35.607, "eval_steps_per_second": 4.451, "step": 57 }, { "epoch": 58.0, "eval_loss": 0.05064238980412483, "eval_runtime": 0.2241, "eval_samples_per_second": 35.694, "eval_steps_per_second": 4.462, "step": 58 }, { "epoch": 59.0, "eval_loss": 0.0496056042611599, "eval_runtime": 0.2358, "eval_samples_per_second": 33.922, "eval_steps_per_second": 4.24, "step": 59 }, { "epoch": 60.0, "eval_loss": 0.04849918186664581, "eval_runtime": 0.234, "eval_samples_per_second": 34.194, "eval_steps_per_second": 4.274, "step": 60 }, { "epoch": 61.0, "eval_loss": 0.04720243811607361, "eval_runtime": 0.2354, "eval_samples_per_second": 33.992, "eval_steps_per_second": 4.249, "step": 61 }, { "epoch": 62.0, "eval_loss": 0.04599544033408165, "eval_runtime": 0.2346, "eval_samples_per_second": 34.103, "eval_steps_per_second": 4.263, "step": 62 }, { "epoch": 63.0, "eval_loss": 0.04451025277376175, "eval_runtime": 0.2405, "eval_samples_per_second": 33.268, "eval_steps_per_second": 4.159, "step": 63 }, { "epoch": 64.0, "eval_loss": 0.04322642832994461, "eval_runtime": 0.2321, "eval_samples_per_second": 34.47, "eval_steps_per_second": 4.309, "step": 64 }, { "epoch": 65.0, "eval_loss": 0.04210076108574867, "eval_runtime": 0.2361, "eval_samples_per_second": 33.885, "eval_steps_per_second": 4.236, "step": 65 }, { "epoch": 66.0, "eval_loss": 0.04086441919207573, "eval_runtime": 0.2341, "eval_samples_per_second": 34.177, "eval_steps_per_second": 4.272, "step": 66 }, { "epoch": 67.0, "eval_loss": 0.03964550420641899, "eval_runtime": 0.2287, "eval_samples_per_second": 34.977, "eval_steps_per_second": 4.372, "step": 67 }, { "epoch": 68.0, "eval_loss": 0.03853166475892067, "eval_runtime": 0.2351, "eval_samples_per_second": 34.031, "eval_steps_per_second": 4.254, "step": 68 }, { "epoch": 69.0, "eval_loss": 0.03747276961803436, "eval_runtime": 0.2354, "eval_samples_per_second": 33.984, "eval_steps_per_second": 4.248, "step": 69 }, { "epoch": 70.0, "eval_loss": 0.0365348644554615, "eval_runtime": 0.2257, "eval_samples_per_second": 35.448, "eval_steps_per_second": 4.431, "step": 70 }, { "epoch": 71.0, "eval_loss": 0.035752058029174805, "eval_runtime": 0.2347, "eval_samples_per_second": 34.085, "eval_steps_per_second": 4.261, "step": 71 }, { "epoch": 72.0, "eval_loss": 0.03502313047647476, "eval_runtime": 0.2313, "eval_samples_per_second": 34.586, "eval_steps_per_second": 4.323, "step": 72 }, { "epoch": 73.0, "eval_loss": 0.03442569449543953, "eval_runtime": 0.2356, "eval_samples_per_second": 33.962, "eval_steps_per_second": 4.245, "step": 73 }, { "epoch": 74.0, "eval_loss": 0.03383530303835869, "eval_runtime": 0.2258, "eval_samples_per_second": 35.436, "eval_steps_per_second": 4.429, "step": 74 }, { "epoch": 75.0, "eval_loss": 0.0333658903837204, "eval_runtime": 0.2352, "eval_samples_per_second": 34.018, "eval_steps_per_second": 4.252, "step": 75 }, { "epoch": 76.0, "eval_loss": 0.03294898569583893, "eval_runtime": 0.2302, "eval_samples_per_second": 34.754, "eval_steps_per_second": 4.344, "step": 76 }, { "epoch": 77.0, "eval_loss": 0.03257429972290993, "eval_runtime": 0.2347, "eval_samples_per_second": 34.09, "eval_steps_per_second": 4.261, "step": 77 }, { "epoch": 78.0, "eval_loss": 0.032144706696271896, "eval_runtime": 0.2353, "eval_samples_per_second": 33.996, "eval_steps_per_second": 4.25, "step": 78 }, { "epoch": 79.0, "eval_loss": 0.03174722567200661, "eval_runtime": 0.2359, "eval_samples_per_second": 33.913, "eval_steps_per_second": 4.239, "step": 79 }, { "epoch": 80.0, "eval_loss": 0.031351201236248016, "eval_runtime": 0.2302, "eval_samples_per_second": 34.75, "eval_steps_per_second": 4.344, "step": 80 }, { "epoch": 81.0, "eval_loss": 0.03098256327211857, "eval_runtime": 0.235, "eval_samples_per_second": 34.037, "eval_steps_per_second": 4.255, "step": 81 }, { "epoch": 82.0, "eval_loss": 0.030596623197197914, "eval_runtime": 0.2313, "eval_samples_per_second": 34.582, "eval_steps_per_second": 4.323, "step": 82 }, { "epoch": 83.0, "eval_loss": 0.0302561167627573, "eval_runtime": 0.2311, "eval_samples_per_second": 34.612, "eval_steps_per_second": 4.326, "step": 83 }, { "epoch": 84.0, "eval_loss": 0.02994276024401188, "eval_runtime": 0.2345, "eval_samples_per_second": 34.114, "eval_steps_per_second": 4.264, "step": 84 }, { "epoch": 85.0, "eval_loss": 0.029655389487743378, "eval_runtime": 0.2317, "eval_samples_per_second": 34.52, "eval_steps_per_second": 4.315, "step": 85 }, { "epoch": 86.0, "eval_loss": 0.029432008042931557, "eval_runtime": 0.2323, "eval_samples_per_second": 34.432, "eval_steps_per_second": 4.304, "step": 86 }, { "epoch": 87.0, "eval_loss": 0.02922823093831539, "eval_runtime": 0.2355, "eval_samples_per_second": 33.969, "eval_steps_per_second": 4.246, "step": 87 }, { "epoch": 88.0, "eval_loss": 0.02902672439813614, "eval_runtime": 0.2356, "eval_samples_per_second": 33.956, "eval_steps_per_second": 4.244, "step": 88 }, { "epoch": 89.0, "eval_loss": 0.02884937822818756, "eval_runtime": 0.2358, "eval_samples_per_second": 33.922, "eval_steps_per_second": 4.24, "step": 89 }, { "epoch": 90.0, "eval_loss": 0.02867262251675129, "eval_runtime": 0.2321, "eval_samples_per_second": 34.471, "eval_steps_per_second": 4.309, "step": 90 }, { "epoch": 91.0, "eval_loss": 0.02851748839020729, "eval_runtime": 0.236, "eval_samples_per_second": 33.898, "eval_steps_per_second": 4.237, "step": 91 }, { "epoch": 92.0, "eval_loss": 0.02836400270462036, "eval_runtime": 0.2256, "eval_samples_per_second": 35.466, "eval_steps_per_second": 4.433, "step": 92 }, { "epoch": 93.0, "eval_loss": 0.028240736573934555, "eval_runtime": 0.2355, "eval_samples_per_second": 33.963, "eval_steps_per_second": 4.245, "step": 93 }, { "epoch": 94.0, "eval_loss": 0.02814488857984543, "eval_runtime": 0.231, "eval_samples_per_second": 34.627, "eval_steps_per_second": 4.328, "step": 94 }, { "epoch": 95.0, "eval_loss": 0.0280709657818079, "eval_runtime": 0.2356, "eval_samples_per_second": 33.953, "eval_steps_per_second": 4.244, "step": 95 }, { "epoch": 96.0, "eval_loss": 0.02801009640097618, "eval_runtime": 0.2315, "eval_samples_per_second": 34.556, "eval_steps_per_second": 4.319, "step": 96 }, { "epoch": 97.0, "eval_loss": 0.02794581837952137, "eval_runtime": 0.2362, "eval_samples_per_second": 33.87, "eval_steps_per_second": 4.234, "step": 97 }, { "epoch": 98.0, "eval_loss": 0.027881627902388573, "eval_runtime": 0.2362, "eval_samples_per_second": 33.875, "eval_steps_per_second": 4.234, "step": 98 }, { "epoch": 99.0, "eval_loss": 0.027840854600071907, "eval_runtime": 0.2356, "eval_samples_per_second": 33.95, "eval_steps_per_second": 4.244, "step": 99 }, { "epoch": 100.0, "learning_rate": 0.0, "loss": 0.8788, "step": 100 }, { "epoch": 100.0, "eval_loss": 0.027821063995361328, "eval_runtime": 0.2225, "eval_samples_per_second": 35.947, "eval_steps_per_second": 4.493, "step": 100 }, { "epoch": 100.0, "step": 100, "total_flos": 487166312448000.0, "train_loss": 0.8787790679931641, "train_runtime": 1000.0694, "train_samples_per_second": 0.8, "train_steps_per_second": 0.1 } ], "logging_steps": 100, "max_steps": 100, "num_train_epochs": 100, "save_steps": 500, "total_flos": 487166312448000.0, "trial_name": null, "trial_params": null }