| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.0, |
| "global_step": 250, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.01, |
| "learning_rate": 6.25e-05, |
| "loss": 32.0931, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.000125, |
| "loss": 21.6268, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001875, |
| "loss": 45.6881, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00025, |
| "loss": 24.4002, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0003125, |
| "loss": 32.4251, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.000375, |
| "loss": 15.8973, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0004375, |
| "loss": 12.2264, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0005, |
| "loss": 18.269, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0004999789344439543, |
| "loss": 24.8883, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0004999157413258782, |
| "loss": 10.1725, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.000499810431295357, |
| "loss": 22.4128, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0004996630220997058, |
| "loss": 8.0502, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0004994735385809777, |
| "loss": 4.9888, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0004992420126717783, |
| "loss": 12.3808, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0004989684833898838, |
| "loss": 5.5916, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0004986529968316653, |
| "loss": 5.0694, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0004982956061643213, |
| "loss": 3.8219, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0004978963716169166, |
| "loss": 7.0727, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0004974553604702333, |
| "loss": 3.9964, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0004969726470454314, |
| "loss": 4.6988, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0004964483126915245, |
| "loss": 6.3211, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0004958824457716707, |
| "loss": 3.4611, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0004952751416482801, |
| "loss": 5.0886, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0004946265026669455, |
| "loss": 5.8484, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.000493936638139193, |
| "loss": 3.6366, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0004932056643240618, |
| "loss": 4.4141, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.000492433704408511, |
| "loss": 3.3416, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.0004916208884866593, |
| "loss": 25.8513, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 0.0004907673535378617, |
| "loss": 5.7299, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 0.0004898732434036243, |
| "loss": 3.0071, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 0.0004889387087633647, |
| "loss": 5.5967, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.0004879639071090174, |
| "loss": 9.4727, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 0.00048694900271849354, |
| "loss": 4.1731, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 0.00048589416662799553, |
| "loss": 5.4418, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 0.00048479957660319374, |
| "loss": 3.4628, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 0.0004836654171092683, |
| "loss": 2.9449, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 0.0004824918792798222, |
| "loss": 12.9026, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 0.00048127916088467097, |
| "loss": 4.0065, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 0.00048002746629651296, |
| "loss": 8.9908, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 0.0004787370064564883, |
| "loss": 14.5672, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 0.0004774079988386296, |
| "loss": 3.6966, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 0.00047604066741321254, |
| "loss": 23.8498, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 0.0004746352426090114, |
| "loss": 21.1258, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 0.00047319196127446593, |
| "loss": 3.9212, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 0.0004717110666377669, |
| "loss": 43.0902, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 0.0004701928082658661, |
| "loss": 8.5586, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 0.00046863744202241806, |
| "loss": 3.2395, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 0.00046704523002466097, |
| "loss": 3.8458, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 0.00046541644059924347, |
| "loss": 19.882, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 0.000463751348237005, |
| "loss": 3.346, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 0.0004620502335467174, |
| "loss": 6.4131, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 0.0004603133832077953, |
| "loss": 3.6067, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 0.00045854108992198417, |
| "loss": 9.6575, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 0.00045673365236403217, |
| "loss": 4.8232, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 0.0004548913751313568, |
| "loss": 3.095, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 0.0004530145686927125, |
| "loss": 2.8664, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 0.00045110354933586896, |
| "loss": 4.9126, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 0.000449158639114309, |
| "loss": 4.293, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 0.0004471801657929551, |
| "loss": 16.1544, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 0.000445168462792932, |
| "loss": 3.619, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 0.0004431238691353784, |
| "loss": 4.2334, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 0.0004410467293843123, |
| "loss": 4.6651, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 0.0004389373935885646, |
| "loss": 2.7114, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 0.0004367962172227866, |
| "loss": 3.0416, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 0.0004346235611275443, |
| "loss": 2.6285, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 0.0004324197914485075, |
| "loss": 3.2878, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 0.0004301852795747458, |
| "loss": 2.2344, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 0.00042792040207614005, |
| "loss": 7.159, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 0.0004256255406399213, |
| "loss": 5.045, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 0.00042330108200634725, |
| "loss": 3.7218, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 0.00042094741790352675, |
| "loss": 2.5571, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 0.0004185649449814045, |
| "loss": 3.4855, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 0.00041615406474491533, |
| "loss": 3.8208, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 0.0004137151834863213, |
| "loss": 4.1545, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 0.000411248712216741, |
| "loss": 3.5336, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 0.0004087550665968846, |
| "loss": 4.7027, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 0.0004062346668670046, |
| "loss": 3.6096, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 0.0004036879377760753, |
| "loss": 3.7848, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 0.0004011153085102116, |
| "loss": 3.2815, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 0.00039851721262034156, |
| "loss": 3.8707, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 0.0003958940879491418, |
| "loss": 3.5825, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 0.00039324637655725053, |
| "loss": 3.2342, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 0.00039057452464876945, |
| "loss": 2.8517, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 0.00038787898249606766, |
| "loss": 3.8141, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 0.0003851602043638994, |
| "loss": 3.3456, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 0.00038241864843284966, |
| "loss": 2.8308, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 0.00037965477672211934, |
| "loss": 3.3316, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 0.0003768690550116639, |
| "loss": 4.581, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 0.00037406195276369793, |
| "loss": 14.9967, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 0.0003712339430435792, |
| "loss": 2.3378, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 0.00036838550244008573, |
| "loss": 3.0834, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 0.00036551711098509903, |
| "loss": 2.462, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 0.0003626292520727067, |
| "loss": 2.9666, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 0.0003597224123777389, |
| "loss": 3.0545, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 0.0003567970817737518, |
| "loss": 2.8931, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 0.00035385375325047166, |
| "loss": 3.9564, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 0.0003508929228307142, |
| "loss": 43.6725, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 0.0003479150894867926, |
| "loss": 4.7077, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 0.0003449207550564285, |
| "loss": 3.0666, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 0.00034191042415818, |
| "loss": 4.0261, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 0.0003388846041064012, |
| "loss": 2.9247, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 0.0003358438048257472, |
| "loss": 2.6719, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 0.0003327885387652391, |
| "loss": 2.6736, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 0.0003297193208119047, |
| "loss": 4.1931, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 0.0003266366682040063, |
| "loss": 2.3891, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 0.0003235411004438741, |
| "loss": 3.2507, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 0.00032043313921035745, |
| "loss": 17.5211, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 0.00031731330827090864, |
| "loss": 2.7262, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 0.0003141821333933158, |
| "loss": 8.2812, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 0.00031104014225709787, |
| "loss": 2.6574, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 0.0003078878643645778, |
| "loss": 2.6401, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 0.00030472583095164874, |
| "loss": 3.0881, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 0.00030155457489824706, |
| "loss": 2.3126, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 0.0002983746306385499, |
| "loss": 3.3111, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 0.00029518653407090953, |
| "loss": 2.9657, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 0.00029199082246754123, |
| "loss": 2.2295, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 0.00028878803438398015, |
| "loss": 3.0558, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 0.00028557870956832135, |
| "loss": 3.4606, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 0.00028236338887025885, |
| "loss": 5.602, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 0.0002791426141499398, |
| "loss": 3.35, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 0.0002759169281866472, |
| "loss": 3.3369, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 0.0002726868745873286, |
| "loss": 4.6327, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 0.00026945299769498494, |
| "loss": 3.6527, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 0.00026621584249693575, |
| "loss": 3.0723, |
| "step": 124 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 0.00026297595453297494, |
| "loss": 3.3674, |
| "step": 125 |
| }, |
| { |
| "epoch": 1.01, |
| "learning_rate": 0.0002597338798034344, |
| "loss": 2.6131, |
| "step": 126 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 0.0002564901646771696, |
| "loss": 2.7034, |
| "step": 127 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 0.00025324535579948274, |
| "loss": 3.2532, |
| "step": 128 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 0.00025, |
| "loss": 2.8019, |
| "step": 129 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 0.00024675464420051733, |
| "loss": 2.1677, |
| "step": 130 |
| }, |
| { |
| "epoch": 1.05, |
| "learning_rate": 0.00024350983532283048, |
| "loss": 2.2027, |
| "step": 131 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 0.0002402661201965656, |
| "loss": 3.0198, |
| "step": 132 |
| }, |
| { |
| "epoch": 1.06, |
| "learning_rate": 0.0002370240454670251, |
| "loss": 3.1579, |
| "step": 133 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 0.00023378415750306423, |
| "loss": 4.3324, |
| "step": 134 |
| }, |
| { |
| "epoch": 1.08, |
| "learning_rate": 0.000230547002305015, |
| "loss": 3.7997, |
| "step": 135 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 0.00022731312541267144, |
| "loss": 2.4695, |
| "step": 136 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 0.00022408307181335285, |
| "loss": 2.729, |
| "step": 137 |
| }, |
| { |
| "epoch": 1.1, |
| "learning_rate": 0.00022085738585006024, |
| "loss": 3.7182, |
| "step": 138 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 0.00021763661112974113, |
| "loss": 2.4506, |
| "step": 139 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 0.00021442129043167875, |
| "loss": 3.4198, |
| "step": 140 |
| }, |
| { |
| "epoch": 1.13, |
| "learning_rate": 0.00021121196561601994, |
| "loss": 2.7518, |
| "step": 141 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 0.00020800917753245875, |
| "loss": 2.756, |
| "step": 142 |
| }, |
| { |
| "epoch": 1.14, |
| "learning_rate": 0.0002048134659290905, |
| "loss": 2.7646, |
| "step": 143 |
| }, |
| { |
| "epoch": 1.15, |
| "learning_rate": 0.0002016253693614501, |
| "loss": 3.6143, |
| "step": 144 |
| }, |
| { |
| "epoch": 1.16, |
| "learning_rate": 0.000198445425101753, |
| "loss": 2.8954, |
| "step": 145 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 0.00019527416904835133, |
| "loss": 2.3892, |
| "step": 146 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 0.00019211213563542217, |
| "loss": 2.7106, |
| "step": 147 |
| }, |
| { |
| "epoch": 1.18, |
| "learning_rate": 0.0001889598577429022, |
| "loss": 2.2542, |
| "step": 148 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 0.00018581786660668432, |
| "loss": 2.7119, |
| "step": 149 |
| }, |
| { |
| "epoch": 1.2, |
| "learning_rate": 0.00018268669172909137, |
| "loss": 2.4768, |
| "step": 150 |
| }, |
| { |
| "epoch": 1.21, |
| "learning_rate": 0.00017956686078964256, |
| "loss": 3.3451, |
| "step": 151 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 0.00017645889955612593, |
| "loss": 54.6038, |
| "step": 152 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 0.00017336333179599378, |
| "loss": 3.2565, |
| "step": 153 |
| }, |
| { |
| "epoch": 1.23, |
| "learning_rate": 0.00017028067918809536, |
| "loss": 2.1254, |
| "step": 154 |
| }, |
| { |
| "epoch": 1.24, |
| "learning_rate": 0.00016721146123476095, |
| "loss": 2.5271, |
| "step": 155 |
| }, |
| { |
| "epoch": 1.25, |
| "learning_rate": 0.00016415619517425294, |
| "loss": 2.5693, |
| "step": 156 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 0.00016111539589359886, |
| "loss": 2.9607, |
| "step": 157 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 0.00015808957584181996, |
| "loss": 2.7638, |
| "step": 158 |
| }, |
| { |
| "epoch": 1.27, |
| "learning_rate": 0.00015507924494357153, |
| "loss": 2.5562, |
| "step": 159 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 0.00015208491051320745, |
| "loss": 2.238, |
| "step": 160 |
| }, |
| { |
| "epoch": 1.29, |
| "learning_rate": 0.00014910707716928588, |
| "loss": 3.1492, |
| "step": 161 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 0.0001461462467495284, |
| "loss": 3.993, |
| "step": 162 |
| }, |
| { |
| "epoch": 1.3, |
| "learning_rate": 0.0001432029182262482, |
| "loss": 2.1614, |
| "step": 163 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 0.00014027758762226108, |
| "loss": 2.6629, |
| "step": 164 |
| }, |
| { |
| "epoch": 1.32, |
| "learning_rate": 0.0001373707479272933, |
| "loss": 2.0814, |
| "step": 165 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 0.00013448288901490095, |
| "loss": 2.454, |
| "step": 166 |
| }, |
| { |
| "epoch": 1.34, |
| "learning_rate": 0.00013161449755991425, |
| "loss": 2.2175, |
| "step": 167 |
| }, |
| { |
| "epoch": 1.34, |
| "learning_rate": 0.00012876605695642086, |
| "loss": 2.2369, |
| "step": 168 |
| }, |
| { |
| "epoch": 1.35, |
| "learning_rate": 0.00012593804723630208, |
| "loss": 1.9271, |
| "step": 169 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 0.0001231309449883361, |
| "loss": 25.7659, |
| "step": 170 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 0.0001203452232778807, |
| "loss": 2.3509, |
| "step": 171 |
| }, |
| { |
| "epoch": 1.38, |
| "learning_rate": 0.00011758135156715042, |
| "loss": 2.2711, |
| "step": 172 |
| }, |
| { |
| "epoch": 1.38, |
| "learning_rate": 0.0001148397956361007, |
| "loss": 1.7004, |
| "step": 173 |
| }, |
| { |
| "epoch": 1.39, |
| "learning_rate": 0.00011212101750393236, |
| "loss": 3.7451, |
| "step": 174 |
| }, |
| { |
| "epoch": 1.4, |
| "learning_rate": 0.00010942547535123056, |
| "loss": 3.5397, |
| "step": 175 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 0.00010675362344274952, |
| "loss": 2.8701, |
| "step": 176 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 0.00010410591205085817, |
| "loss": 10.4424, |
| "step": 177 |
| }, |
| { |
| "epoch": 1.42, |
| "learning_rate": 0.00010148278737965844, |
| "loss": 2.1697, |
| "step": 178 |
| }, |
| { |
| "epoch": 1.43, |
| "learning_rate": 9.888469148978843e-05, |
| "loss": 2.487, |
| "step": 179 |
| }, |
| { |
| "epoch": 1.44, |
| "learning_rate": 9.63120622239248e-05, |
| "loss": 2.6825, |
| "step": 180 |
| }, |
| { |
| "epoch": 1.45, |
| "learning_rate": 9.376533313299543e-05, |
| "loss": 2.043, |
| "step": 181 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 9.124493340311537e-05, |
| "loss": 5.7818, |
| "step": 182 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 8.875128778325903e-05, |
| "loss": 2.9812, |
| "step": 183 |
| }, |
| { |
| "epoch": 1.47, |
| "learning_rate": 8.628481651367875e-05, |
| "loss": 3.0498, |
| "step": 184 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 8.384593525508466e-05, |
| "loss": 2.1369, |
| "step": 185 |
| }, |
| { |
| "epoch": 1.49, |
| "learning_rate": 8.143505501859552e-05, |
| "loss": 2.6704, |
| "step": 186 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 7.905258209647325e-05, |
| "loss": 17.568, |
| "step": 187 |
| }, |
| { |
| "epoch": 1.5, |
| "learning_rate": 7.669891799365283e-05, |
| "loss": 3.1703, |
| "step": 188 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 7.437445936007866e-05, |
| "loss": 3.0861, |
| "step": 189 |
| }, |
| { |
| "epoch": 1.52, |
| "learning_rate": 7.207959792385999e-05, |
| "loss": 3.5536, |
| "step": 190 |
| }, |
| { |
| "epoch": 1.53, |
| "learning_rate": 6.981472042525417e-05, |
| "loss": 2.0992, |
| "step": 191 |
| }, |
| { |
| "epoch": 1.54, |
| "learning_rate": 6.75802085514925e-05, |
| "loss": 2.9718, |
| "step": 192 |
| }, |
| { |
| "epoch": 1.54, |
| "learning_rate": 6.537643887245573e-05, |
| "loss": 2.4697, |
| "step": 193 |
| }, |
| { |
| "epoch": 1.55, |
| "learning_rate": 6.320378277721342e-05, |
| "loss": 29.7439, |
| "step": 194 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 6.106260641143546e-05, |
| "loss": 1.9889, |
| "step": 195 |
| }, |
| { |
| "epoch": 1.57, |
| "learning_rate": 5.895327061568775e-05, |
| "loss": 4.2758, |
| "step": 196 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 5.6876130864621706e-05, |
| "loss": 12.9452, |
| "step": 197 |
| }, |
| { |
| "epoch": 1.58, |
| "learning_rate": 5.483153720706799e-05, |
| "loss": 3.6122, |
| "step": 198 |
| }, |
| { |
| "epoch": 1.59, |
| "learning_rate": 5.281983420704498e-05, |
| "loss": 2.7796, |
| "step": 199 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 5.0841360885691e-05, |
| "loss": 2.8641, |
| "step": 200 |
| }, |
| { |
| "epoch": 1.61, |
| "learning_rate": 4.889645066413112e-05, |
| "loss": 2.9809, |
| "step": 201 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 4.698543130728755e-05, |
| "loss": 19.311, |
| "step": 202 |
| }, |
| { |
| "epoch": 1.62, |
| "learning_rate": 4.510862486864317e-05, |
| "loss": 2.1423, |
| "step": 203 |
| }, |
| { |
| "epoch": 1.63, |
| "learning_rate": 4.3266347635967846e-05, |
| "loss": 3.2833, |
| "step": 204 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 4.1458910078015886e-05, |
| "loss": 2.2377, |
| "step": 205 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 3.968661679220467e-05, |
| "loss": 13.9436, |
| "step": 206 |
| }, |
| { |
| "epoch": 1.66, |
| "learning_rate": 3.7949766453282656e-05, |
| "loss": 3.2007, |
| "step": 207 |
| }, |
| { |
| "epoch": 1.66, |
| "learning_rate": 3.6248651762995e-05, |
| "loss": 3.3191, |
| "step": 208 |
| }, |
| { |
| "epoch": 1.67, |
| "learning_rate": 3.458355940075653e-05, |
| "loss": 3.3573, |
| "step": 209 |
| }, |
| { |
| "epoch": 1.68, |
| "learning_rate": 3.295476997533905e-05, |
| "loss": 2.4375, |
| "step": 210 |
| }, |
| { |
| "epoch": 1.69, |
| "learning_rate": 3.1362557977582e-05, |
| "loss": 2.6514, |
| "step": 211 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 2.980719173413396e-05, |
| "loss": 2.496, |
| "step": 212 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 2.828893336223315e-05, |
| "loss": 2.0251, |
| "step": 213 |
| }, |
| { |
| "epoch": 1.71, |
| "learning_rate": 2.6808038725534083e-05, |
| "loss": 3.1729, |
| "step": 214 |
| }, |
| { |
| "epoch": 1.72, |
| "learning_rate": 2.536475739098862e-05, |
| "loss": 3.7772, |
| "step": 215 |
| }, |
| { |
| "epoch": 1.73, |
| "learning_rate": 2.395933258678745e-05, |
| "loss": 2.2552, |
| "step": 216 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 2.259200116137039e-05, |
| "loss": 3.0183, |
| "step": 217 |
| }, |
| { |
| "epoch": 1.74, |
| "learning_rate": 2.1262993543511715e-05, |
| "loss": 2.2062, |
| "step": 218 |
| }, |
| { |
| "epoch": 1.75, |
| "learning_rate": 1.9972533703487035e-05, |
| "loss": 3.2969, |
| "step": 219 |
| }, |
| { |
| "epoch": 1.76, |
| "learning_rate": 1.872083911532907e-05, |
| "loss": 2.9938, |
| "step": 220 |
| }, |
| { |
| "epoch": 1.77, |
| "learning_rate": 1.7508120720177794e-05, |
| "loss": 2.7953, |
| "step": 221 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 1.6334582890731697e-05, |
| "loss": 1.6648, |
| "step": 222 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 1.5200423396806273e-05, |
| "loss": 3.0712, |
| "step": 223 |
| }, |
| { |
| "epoch": 1.79, |
| "learning_rate": 1.4105833372004522e-05, |
| "loss": 2.1111, |
| "step": 224 |
| }, |
| { |
| "epoch": 1.8, |
| "learning_rate": 1.305099728150655e-05, |
| "loss": 3.0564, |
| "step": 225 |
| }, |
| { |
| "epoch": 1.81, |
| "learning_rate": 1.203609289098262e-05, |
| "loss": 2.968, |
| "step": 226 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 1.1061291236635295e-05, |
| "loss": 3.1569, |
| "step": 227 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 1.0126756596375685e-05, |
| "loss": 2.7812, |
| "step": 228 |
| }, |
| { |
| "epoch": 1.83, |
| "learning_rate": 9.232646462138423e-06, |
| "loss": 2.5548, |
| "step": 229 |
| }, |
| { |
| "epoch": 1.84, |
| "learning_rate": 8.379111513340754e-06, |
| "loss": 2.707, |
| "step": 230 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 7.566295591489053e-06, |
| "loss": 2.7071, |
| "step": 231 |
| }, |
| { |
| "epoch": 1.86, |
| "learning_rate": 6.794335675938179e-06, |
| "loss": 3.0449, |
| "step": 232 |
| }, |
| { |
| "epoch": 1.86, |
| "learning_rate": 6.063361860807026e-06, |
| "loss": 3.2289, |
| "step": 233 |
| }, |
| { |
| "epoch": 1.87, |
| "learning_rate": 5.3734973330546165e-06, |
| "loss": 2.1801, |
| "step": 234 |
| }, |
| { |
| "epoch": 1.88, |
| "learning_rate": 4.7248583517199014e-06, |
| "loss": 1.9365, |
| "step": 235 |
| }, |
| { |
| "epoch": 1.89, |
| "learning_rate": 4.117554228329407e-06, |
| "loss": 3.7057, |
| "step": 236 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 3.551687308475521e-06, |
| "loss": 2.9565, |
| "step": 237 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 3.027352954568713e-06, |
| "loss": 2.0448, |
| "step": 238 |
| }, |
| { |
| "epoch": 1.91, |
| "learning_rate": 2.544639529766829e-06, |
| "loss": 4.4805, |
| "step": 239 |
| }, |
| { |
| "epoch": 1.92, |
| "learning_rate": 2.1036283830834224e-06, |
| "loss": 2.2209, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.93, |
| "learning_rate": 1.704393835678747e-06, |
| "loss": 1.9931, |
| "step": 241 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.347003168334665e-06, |
| "loss": 2.951, |
| "step": 242 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.0315166101162454e-06, |
| "loss": 3.6584, |
| "step": 243 |
| }, |
| { |
| "epoch": 1.95, |
| "learning_rate": 7.579873282216598e-07, |
| "loss": 4.0625, |
| "step": 244 |
| }, |
| { |
| "epoch": 1.96, |
| "learning_rate": 5.264614190222949e-07, |
| "loss": 2.6563, |
| "step": 245 |
| }, |
| { |
| "epoch": 1.97, |
| "learning_rate": 3.3697790029424415e-07, |
| "loss": 1.9273, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 1.8956870464298393e-07, |
| "loss": 2.0326, |
| "step": 247 |
| }, |
| { |
| "epoch": 1.98, |
| "learning_rate": 8.425867412190091e-08, |
| "loss": 2.1434, |
| "step": 248 |
| }, |
| { |
| "epoch": 1.99, |
| "learning_rate": 2.1065556045779532e-08, |
| "loss": 3.4596, |
| "step": 249 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 0.0, |
| "loss": 2.4318, |
| "step": 250 |
| }, |
| { |
| "epoch": 2.0, |
| "step": 250, |
| "total_flos": 7752148241154048.0, |
| "train_loss": 5.8172749667167665, |
| "train_runtime": 15461.0801, |
| "train_samples_per_second": 0.065, |
| "train_steps_per_second": 0.016 |
| } |
| ], |
| "max_steps": 250, |
| "num_train_epochs": 2, |
| "total_flos": 7752148241154048.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
