| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.8149190710767065, | |
| "eval_steps": 500, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0028149190710767065, | |
| "grad_norm": 0.3785315454006195, | |
| "learning_rate": 4e-05, | |
| "loss": 2.7534, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005629838142153413, | |
| "grad_norm": 0.3462165296077728, | |
| "learning_rate": 8e-05, | |
| "loss": 2.7611, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00844475721323012, | |
| "grad_norm": 0.41192108392715454, | |
| "learning_rate": 0.00012, | |
| "loss": 2.9121, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.011259676284306826, | |
| "grad_norm": 0.4809342324733734, | |
| "learning_rate": 0.00016, | |
| "loss": 2.9178, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.014074595355383532, | |
| "grad_norm": 0.4055400490760803, | |
| "learning_rate": 0.0002, | |
| "loss": 2.9431, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01688951442646024, | |
| "grad_norm": 0.4463144838809967, | |
| "learning_rate": 0.00019979899497487438, | |
| "loss": 2.7718, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.019704433497536946, | |
| "grad_norm": 0.49435216188430786, | |
| "learning_rate": 0.00019959798994974876, | |
| "loss": 2.7197, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.022519352568613652, | |
| "grad_norm": 0.5844013690948486, | |
| "learning_rate": 0.00019939698492462313, | |
| "loss": 2.7076, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.025334271639690358, | |
| "grad_norm": 0.6181543469429016, | |
| "learning_rate": 0.0001991959798994975, | |
| "loss": 2.8055, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.028149190710767064, | |
| "grad_norm": 0.7984749674797058, | |
| "learning_rate": 0.00019899497487437187, | |
| "loss": 2.7629, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03096410978184377, | |
| "grad_norm": 0.6961840391159058, | |
| "learning_rate": 0.00019879396984924622, | |
| "loss": 2.6917, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03377902885292048, | |
| "grad_norm": 0.837709367275238, | |
| "learning_rate": 0.00019859296482412062, | |
| "loss": 2.6108, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.036593947923997186, | |
| "grad_norm": 0.7435119152069092, | |
| "learning_rate": 0.000198391959798995, | |
| "loss": 2.7828, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03940886699507389, | |
| "grad_norm": 0.8047707080841064, | |
| "learning_rate": 0.00019819095477386937, | |
| "loss": 2.7011, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0422237860661506, | |
| "grad_norm": 0.9793757796287537, | |
| "learning_rate": 0.0001979899497487437, | |
| "loss": 2.865, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.045038705137227304, | |
| "grad_norm": 0.7536874413490295, | |
| "learning_rate": 0.0001977889447236181, | |
| "loss": 2.7922, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04785362420830401, | |
| "grad_norm": 0.6820270419120789, | |
| "learning_rate": 0.00019758793969849249, | |
| "loss": 3.1034, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.050668543279380716, | |
| "grad_norm": 0.7651283144950867, | |
| "learning_rate": 0.00019738693467336683, | |
| "loss": 2.8736, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.05348346235045742, | |
| "grad_norm": 0.5590704083442688, | |
| "learning_rate": 0.0001971859296482412, | |
| "loss": 2.6802, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.05629838142153413, | |
| "grad_norm": 0.5996040105819702, | |
| "learning_rate": 0.0001969849246231156, | |
| "loss": 2.6352, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.059113300492610835, | |
| "grad_norm": 0.6097638607025146, | |
| "learning_rate": 0.00019678391959798995, | |
| "loss": 2.5763, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06192821956368754, | |
| "grad_norm": 0.5201358795166016, | |
| "learning_rate": 0.00019658291457286432, | |
| "loss": 2.6492, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.06474313863476425, | |
| "grad_norm": 0.8090603351593018, | |
| "learning_rate": 0.0001963819095477387, | |
| "loss": 2.5719, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.06755805770584096, | |
| "grad_norm": 0.6470005512237549, | |
| "learning_rate": 0.0001961809045226131, | |
| "loss": 2.9084, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07037297677691766, | |
| "grad_norm": 0.6126617193222046, | |
| "learning_rate": 0.00019597989949748744, | |
| "loss": 2.7863, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07318789584799437, | |
| "grad_norm": 0.5378536581993103, | |
| "learning_rate": 0.00019577889447236181, | |
| "loss": 2.6437, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.07600281491907107, | |
| "grad_norm": 0.6851357817649841, | |
| "learning_rate": 0.0001955778894472362, | |
| "loss": 2.6539, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.07881773399014778, | |
| "grad_norm": 0.6153799295425415, | |
| "learning_rate": 0.00019537688442211056, | |
| "loss": 2.67, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08163265306122448, | |
| "grad_norm": 0.5324752926826477, | |
| "learning_rate": 0.00019517587939698493, | |
| "loss": 2.6674, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0844475721323012, | |
| "grad_norm": 0.5797032713890076, | |
| "learning_rate": 0.0001949748743718593, | |
| "loss": 2.5109, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08726249120337791, | |
| "grad_norm": 0.6937679052352905, | |
| "learning_rate": 0.00019477386934673368, | |
| "loss": 2.7514, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.09007741027445461, | |
| "grad_norm": 0.6234177350997925, | |
| "learning_rate": 0.00019457286432160805, | |
| "loss": 2.6147, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09289232934553132, | |
| "grad_norm": 0.6435564756393433, | |
| "learning_rate": 0.00019437185929648243, | |
| "loss": 2.6747, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.09570724841660802, | |
| "grad_norm": 0.6985692381858826, | |
| "learning_rate": 0.0001941708542713568, | |
| "loss": 2.7223, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.09852216748768473, | |
| "grad_norm": 0.6003565788269043, | |
| "learning_rate": 0.00019396984924623117, | |
| "loss": 2.6478, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.10133708655876143, | |
| "grad_norm": 0.7325728535652161, | |
| "learning_rate": 0.00019376884422110552, | |
| "loss": 2.855, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.10415200562983815, | |
| "grad_norm": 0.6490616798400879, | |
| "learning_rate": 0.00019356783919597992, | |
| "loss": 2.8664, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.10696692470091484, | |
| "grad_norm": 0.6137815713882446, | |
| "learning_rate": 0.0001933668341708543, | |
| "loss": 2.5355, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.10978184377199156, | |
| "grad_norm": 0.6218917369842529, | |
| "learning_rate": 0.00019316582914572864, | |
| "loss": 2.7318, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.11259676284306826, | |
| "grad_norm": 0.6341124773025513, | |
| "learning_rate": 0.000192964824120603, | |
| "loss": 2.5333, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11541168191414497, | |
| "grad_norm": 0.5556070804595947, | |
| "learning_rate": 0.0001927638190954774, | |
| "loss": 2.9166, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.11822660098522167, | |
| "grad_norm": 0.5476509928703308, | |
| "learning_rate": 0.00019256281407035178, | |
| "loss": 2.8262, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.12104152005629838, | |
| "grad_norm": 0.8177505731582642, | |
| "learning_rate": 0.00019236180904522613, | |
| "loss": 2.9477, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.12385643912737508, | |
| "grad_norm": 0.6593706011772156, | |
| "learning_rate": 0.0001921608040201005, | |
| "loss": 2.8449, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1266713581984518, | |
| "grad_norm": 0.59237140417099, | |
| "learning_rate": 0.0001919597989949749, | |
| "loss": 2.6079, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1294862772695285, | |
| "grad_norm": 0.5167338252067566, | |
| "learning_rate": 0.00019175879396984925, | |
| "loss": 2.6834, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.13230119634060522, | |
| "grad_norm": 0.5484845042228699, | |
| "learning_rate": 0.00019155778894472362, | |
| "loss": 2.5793, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.13511611541168192, | |
| "grad_norm": 0.5930073261260986, | |
| "learning_rate": 0.000191356783919598, | |
| "loss": 2.757, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.13793103448275862, | |
| "grad_norm": 0.6741965413093567, | |
| "learning_rate": 0.0001911557788944724, | |
| "loss": 2.7182, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.14074595355383532, | |
| "grad_norm": 0.558120608329773, | |
| "learning_rate": 0.00019095477386934674, | |
| "loss": 2.6401, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14356087262491204, | |
| "grad_norm": 0.6161705255508423, | |
| "learning_rate": 0.0001907537688442211, | |
| "loss": 2.6181, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.14637579169598874, | |
| "grad_norm": 0.6661592721939087, | |
| "learning_rate": 0.00019055276381909548, | |
| "loss": 2.7207, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.14919071076706544, | |
| "grad_norm": 0.5285555720329285, | |
| "learning_rate": 0.00019035175879396986, | |
| "loss": 2.5631, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.15200562983814214, | |
| "grad_norm": 0.6050645709037781, | |
| "learning_rate": 0.00019015075376884423, | |
| "loss": 2.4716, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.15482054890921887, | |
| "grad_norm": 0.6041057109832764, | |
| "learning_rate": 0.0001899497487437186, | |
| "loss": 2.7241, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.15763546798029557, | |
| "grad_norm": 0.6147128343582153, | |
| "learning_rate": 0.00018974874371859298, | |
| "loss": 2.6362, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.16045038705137227, | |
| "grad_norm": 0.5417614579200745, | |
| "learning_rate": 0.00018954773869346732, | |
| "loss": 2.8325, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.16326530612244897, | |
| "grad_norm": 0.5944551229476929, | |
| "learning_rate": 0.00018934673366834172, | |
| "loss": 2.6793, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.1660802251935257, | |
| "grad_norm": 0.6394937634468079, | |
| "learning_rate": 0.0001891457286432161, | |
| "loss": 2.721, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1688951442646024, | |
| "grad_norm": 0.5581662058830261, | |
| "learning_rate": 0.00018894472361809047, | |
| "loss": 2.6786, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1717100633356791, | |
| "grad_norm": 0.5921449065208435, | |
| "learning_rate": 0.00018874371859296481, | |
| "loss": 2.5239, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.17452498240675582, | |
| "grad_norm": 0.5900184512138367, | |
| "learning_rate": 0.00018854271356783921, | |
| "loss": 2.6708, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.17733990147783252, | |
| "grad_norm": 0.6194185614585876, | |
| "learning_rate": 0.0001883417085427136, | |
| "loss": 2.9428, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.18015482054890922, | |
| "grad_norm": 0.629349410533905, | |
| "learning_rate": 0.00018814070351758793, | |
| "loss": 2.6705, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.18296973961998592, | |
| "grad_norm": 0.5497152805328369, | |
| "learning_rate": 0.0001879396984924623, | |
| "loss": 2.8205, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.18578465869106264, | |
| "grad_norm": 0.5276259779930115, | |
| "learning_rate": 0.0001877386934673367, | |
| "loss": 2.5922, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.18859957776213934, | |
| "grad_norm": 0.7193230390548706, | |
| "learning_rate": 0.00018753768844221108, | |
| "loss": 2.867, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.19141449683321604, | |
| "grad_norm": 0.6483210325241089, | |
| "learning_rate": 0.00018733668341708543, | |
| "loss": 2.9455, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.19422941590429274, | |
| "grad_norm": 0.7181980013847351, | |
| "learning_rate": 0.0001871356783919598, | |
| "loss": 2.7443, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.19704433497536947, | |
| "grad_norm": 0.6001389026641846, | |
| "learning_rate": 0.0001869346733668342, | |
| "loss": 2.6464, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.19985925404644617, | |
| "grad_norm": 0.7344582080841064, | |
| "learning_rate": 0.00018673366834170854, | |
| "loss": 2.6694, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.20267417311752287, | |
| "grad_norm": 0.6493490934371948, | |
| "learning_rate": 0.00018653266331658292, | |
| "loss": 2.6506, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.20548909218859956, | |
| "grad_norm": 0.5350422859191895, | |
| "learning_rate": 0.0001863316582914573, | |
| "loss": 2.6709, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2083040112596763, | |
| "grad_norm": 0.5754289031028748, | |
| "learning_rate": 0.0001861306532663317, | |
| "loss": 2.3618, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.211118930330753, | |
| "grad_norm": 0.6207188367843628, | |
| "learning_rate": 0.00018592964824120604, | |
| "loss": 2.7498, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.2139338494018297, | |
| "grad_norm": 0.5524656176567078, | |
| "learning_rate": 0.0001857286432160804, | |
| "loss": 2.4996, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.21674876847290642, | |
| "grad_norm": 0.7466227412223816, | |
| "learning_rate": 0.00018552763819095478, | |
| "loss": 2.944, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.21956368754398312, | |
| "grad_norm": 0.6438124179840088, | |
| "learning_rate": 0.00018532663316582915, | |
| "loss": 2.7136, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.22237860661505982, | |
| "grad_norm": 0.6562415957450867, | |
| "learning_rate": 0.00018512562814070353, | |
| "loss": 2.797, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.22519352568613651, | |
| "grad_norm": 0.6599562168121338, | |
| "learning_rate": 0.0001849246231155779, | |
| "loss": 2.511, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.22800844475721324, | |
| "grad_norm": 0.6012830138206482, | |
| "learning_rate": 0.00018472361809045227, | |
| "loss": 2.5117, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.23082336382828994, | |
| "grad_norm": 0.5618470907211304, | |
| "learning_rate": 0.00018452261306532662, | |
| "loss": 2.7258, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.23363828289936664, | |
| "grad_norm": 0.7711282968521118, | |
| "learning_rate": 0.00018432160804020102, | |
| "loss": 2.8518, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.23645320197044334, | |
| "grad_norm": 0.5676078200340271, | |
| "learning_rate": 0.0001841206030150754, | |
| "loss": 2.7253, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.23926812104152007, | |
| "grad_norm": 0.8567176461219788, | |
| "learning_rate": 0.00018391959798994977, | |
| "loss": 2.767, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.24208304011259676, | |
| "grad_norm": 0.5816414952278137, | |
| "learning_rate": 0.0001837185929648241, | |
| "loss": 2.5211, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.24489795918367346, | |
| "grad_norm": 0.5357186198234558, | |
| "learning_rate": 0.0001835175879396985, | |
| "loss": 2.5882, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.24771287825475016, | |
| "grad_norm": 0.5406627655029297, | |
| "learning_rate": 0.00018331658291457288, | |
| "loss": 2.3529, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.25052779732582686, | |
| "grad_norm": 0.9183681607246399, | |
| "learning_rate": 0.00018311557788944723, | |
| "loss": 2.9405, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2533427163969036, | |
| "grad_norm": 0.5938777327537537, | |
| "learning_rate": 0.0001829145728643216, | |
| "loss": 2.8762, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2561576354679803, | |
| "grad_norm": 0.559532880783081, | |
| "learning_rate": 0.000182713567839196, | |
| "loss": 2.4396, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.258972554539057, | |
| "grad_norm": 0.8062023520469666, | |
| "learning_rate": 0.00018251256281407038, | |
| "loss": 2.8835, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2617874736101337, | |
| "grad_norm": 0.5407679080963135, | |
| "learning_rate": 0.00018231155778894472, | |
| "loss": 2.5692, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.26460239268121044, | |
| "grad_norm": 0.5537972450256348, | |
| "learning_rate": 0.0001821105527638191, | |
| "loss": 2.4051, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2674173117522871, | |
| "grad_norm": 0.6128715872764587, | |
| "learning_rate": 0.0001819095477386935, | |
| "loss": 2.6613, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.27023223082336384, | |
| "grad_norm": 0.9666823148727417, | |
| "learning_rate": 0.00018170854271356784, | |
| "loss": 2.6943, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.2730471498944405, | |
| "grad_norm": 0.587451696395874, | |
| "learning_rate": 0.00018150753768844221, | |
| "loss": 2.5654, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.27586206896551724, | |
| "grad_norm": 0.6436663269996643, | |
| "learning_rate": 0.0001813065326633166, | |
| "loss": 2.5957, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.27867698803659396, | |
| "grad_norm": 0.5880750417709351, | |
| "learning_rate": 0.00018110552763819096, | |
| "loss": 2.8445, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.28149190710767064, | |
| "grad_norm": 0.5972994565963745, | |
| "learning_rate": 0.00018090452261306533, | |
| "loss": 2.6186, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.28430682617874736, | |
| "grad_norm": 0.5434820652008057, | |
| "learning_rate": 0.0001807035175879397, | |
| "loss": 2.8376, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.2871217452498241, | |
| "grad_norm": 0.5735207200050354, | |
| "learning_rate": 0.00018050251256281408, | |
| "loss": 2.4118, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.28993666432090076, | |
| "grad_norm": 0.5313388705253601, | |
| "learning_rate": 0.00018030150753768845, | |
| "loss": 2.468, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.2927515833919775, | |
| "grad_norm": 0.6161223649978638, | |
| "learning_rate": 0.00018010050251256282, | |
| "loss": 2.663, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.2955665024630542, | |
| "grad_norm": 0.5644655227661133, | |
| "learning_rate": 0.0001798994974874372, | |
| "loss": 2.3672, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.2983814215341309, | |
| "grad_norm": 0.6080154776573181, | |
| "learning_rate": 0.00017969849246231157, | |
| "loss": 2.6672, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3011963406052076, | |
| "grad_norm": 0.5323423147201538, | |
| "learning_rate": 0.00017949748743718592, | |
| "loss": 2.8084, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.3040112596762843, | |
| "grad_norm": 0.5441535711288452, | |
| "learning_rate": 0.00017929648241206032, | |
| "loss": 2.5269, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.306826178747361, | |
| "grad_norm": 0.5068178772926331, | |
| "learning_rate": 0.0001790954773869347, | |
| "loss": 2.5472, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.30964109781843774, | |
| "grad_norm": 0.6056650876998901, | |
| "learning_rate": 0.00017889447236180906, | |
| "loss": 2.6083, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3124560168895144, | |
| "grad_norm": 0.5633851885795593, | |
| "learning_rate": 0.0001786934673366834, | |
| "loss": 2.5353, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.31527093596059114, | |
| "grad_norm": 0.6467467546463013, | |
| "learning_rate": 0.0001784924623115578, | |
| "loss": 2.7402, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.31808585503166786, | |
| "grad_norm": 0.590074360370636, | |
| "learning_rate": 0.00017829145728643218, | |
| "loss": 2.7417, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.32090077410274453, | |
| "grad_norm": 0.5952100157737732, | |
| "learning_rate": 0.00017809045226130653, | |
| "loss": 2.4225, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.32371569317382126, | |
| "grad_norm": 0.5567030310630798, | |
| "learning_rate": 0.0001778894472361809, | |
| "loss": 2.5335, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.32653061224489793, | |
| "grad_norm": 0.6068913340568542, | |
| "learning_rate": 0.0001776884422110553, | |
| "loss": 2.6689, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.32934553131597466, | |
| "grad_norm": 0.5481736660003662, | |
| "learning_rate": 0.00017748743718592967, | |
| "loss": 2.5618, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.3321604503870514, | |
| "grad_norm": 0.5849531888961792, | |
| "learning_rate": 0.00017728643216080402, | |
| "loss": 2.8453, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.33497536945812806, | |
| "grad_norm": 0.6313461065292358, | |
| "learning_rate": 0.0001770854271356784, | |
| "loss": 2.6883, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.3377902885292048, | |
| "grad_norm": 1.3009490966796875, | |
| "learning_rate": 0.0001768844221105528, | |
| "loss": 2.5748, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3406052076002815, | |
| "grad_norm": 0.5591140985488892, | |
| "learning_rate": 0.00017668341708542714, | |
| "loss": 2.7251, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.3434201266713582, | |
| "grad_norm": 0.5167106986045837, | |
| "learning_rate": 0.0001764824120603015, | |
| "loss": 2.6753, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.3462350457424349, | |
| "grad_norm": 0.6912369132041931, | |
| "learning_rate": 0.00017628140703517588, | |
| "loss": 2.9036, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.34904996481351164, | |
| "grad_norm": 0.7323533892631531, | |
| "learning_rate": 0.00017608040201005026, | |
| "loss": 2.7168, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.3518648838845883, | |
| "grad_norm": 0.5843552947044373, | |
| "learning_rate": 0.00017587939698492463, | |
| "loss": 2.5335, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.35467980295566504, | |
| "grad_norm": 0.6250912547111511, | |
| "learning_rate": 0.000175678391959799, | |
| "loss": 2.7493, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.3574947220267417, | |
| "grad_norm": 0.5447134375572205, | |
| "learning_rate": 0.00017547738693467338, | |
| "loss": 2.5758, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.36030964109781843, | |
| "grad_norm": 0.8142397403717041, | |
| "learning_rate": 0.00017527638190954775, | |
| "loss": 2.8021, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.36312456016889516, | |
| "grad_norm": NaN, | |
| "learning_rate": 0.00017527638190954775, | |
| "loss": 2.6983, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.36593947923997183, | |
| "grad_norm": 0.5528063774108887, | |
| "learning_rate": 0.00017507537688442212, | |
| "loss": 2.695, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.36875439831104856, | |
| "grad_norm": 0.605383574962616, | |
| "learning_rate": 0.0001748743718592965, | |
| "loss": 2.7013, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.3715693173821253, | |
| "grad_norm": 0.627310037612915, | |
| "learning_rate": 0.00017467336683417087, | |
| "loss": 2.7744, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.37438423645320196, | |
| "grad_norm": 0.6117985844612122, | |
| "learning_rate": 0.00017447236180904521, | |
| "loss": 2.9001, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.3771991555242787, | |
| "grad_norm": 0.5570118427276611, | |
| "learning_rate": 0.00017427135678391961, | |
| "loss": 2.6795, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3800140745953554, | |
| "grad_norm": 0.6382287740707397, | |
| "learning_rate": 0.000174070351758794, | |
| "loss": 2.8177, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3828289936664321, | |
| "grad_norm": 0.7003315091133118, | |
| "learning_rate": 0.00017386934673366836, | |
| "loss": 2.531, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.3856439127375088, | |
| "grad_norm": 0.5270616412162781, | |
| "learning_rate": 0.0001736683417085427, | |
| "loss": 2.7267, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.3884588318085855, | |
| "grad_norm": 0.6856080889701843, | |
| "learning_rate": 0.0001734673366834171, | |
| "loss": 2.8481, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.3912737508796622, | |
| "grad_norm": 0.7767484784126282, | |
| "learning_rate": 0.00017326633165829148, | |
| "loss": 2.7403, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.39408866995073893, | |
| "grad_norm": 0.5755979418754578, | |
| "learning_rate": 0.00017306532663316582, | |
| "loss": 2.6526, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3969035890218156, | |
| "grad_norm": 0.6108975410461426, | |
| "learning_rate": 0.0001728643216080402, | |
| "loss": 2.5531, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.39971850809289233, | |
| "grad_norm": 0.6080026030540466, | |
| "learning_rate": 0.0001726633165829146, | |
| "loss": 2.7344, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.40253342716396906, | |
| "grad_norm": 0.5954862833023071, | |
| "learning_rate": 0.00017246231155778897, | |
| "loss": 2.6179, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.40534834623504573, | |
| "grad_norm": 0.7604647874832153, | |
| "learning_rate": 0.00017226130653266332, | |
| "loss": 2.8556, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 0.540407657623291, | |
| "learning_rate": 0.0001720603015075377, | |
| "loss": 2.671, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.41097818437719913, | |
| "grad_norm": 0.5598605871200562, | |
| "learning_rate": 0.00017185929648241206, | |
| "loss": 2.697, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.41379310344827586, | |
| "grad_norm": 0.6206982135772705, | |
| "learning_rate": 0.00017165829145728644, | |
| "loss": 2.5687, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.4166080225193526, | |
| "grad_norm": 0.5486766695976257, | |
| "learning_rate": 0.0001714572864321608, | |
| "loss": 2.8242, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.41942294159042925, | |
| "grad_norm": 0.7132663130760193, | |
| "learning_rate": 0.00017125628140703518, | |
| "loss": 2.8706, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.422237860661506, | |
| "grad_norm": 0.6359018087387085, | |
| "learning_rate": 0.00017105527638190955, | |
| "loss": 2.773, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4250527797325827, | |
| "grad_norm": 0.5943129062652588, | |
| "learning_rate": 0.00017085427135678393, | |
| "loss": 2.6535, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.4278676988036594, | |
| "grad_norm": 0.6567736864089966, | |
| "learning_rate": 0.0001706532663316583, | |
| "loss": 2.6059, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.4306826178747361, | |
| "grad_norm": 0.6345821619033813, | |
| "learning_rate": 0.00017045226130653267, | |
| "loss": 2.5776, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.43349753694581283, | |
| "grad_norm": 0.9386352896690369, | |
| "learning_rate": 0.00017025125628140705, | |
| "loss": 2.7449, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.4363124560168895, | |
| "grad_norm": 0.5455414652824402, | |
| "learning_rate": 0.00017005025125628142, | |
| "loss": 2.3967, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.43912737508796623, | |
| "grad_norm": 0.7040349841117859, | |
| "learning_rate": 0.0001698492462311558, | |
| "loss": 2.9214, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.4419422941590429, | |
| "grad_norm": 0.5507174730300903, | |
| "learning_rate": 0.00016964824120603016, | |
| "loss": 2.5087, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.44475721323011963, | |
| "grad_norm": 0.6239134669303894, | |
| "learning_rate": 0.0001694472361809045, | |
| "loss": 3.2032, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.44757213230119636, | |
| "grad_norm": 0.7403885722160339, | |
| "learning_rate": 0.0001692462311557789, | |
| "loss": 2.8829, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.45038705137227303, | |
| "grad_norm": 0.5260657072067261, | |
| "learning_rate": 0.00016904522613065328, | |
| "loss": 2.4572, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.45320197044334976, | |
| "grad_norm": 0.7505115270614624, | |
| "learning_rate": 0.00016884422110552766, | |
| "loss": 2.7264, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.4560168895144265, | |
| "grad_norm": 0.6088585257530212, | |
| "learning_rate": 0.000168643216080402, | |
| "loss": 2.5714, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.45883180858550315, | |
| "grad_norm": 0.6011828184127808, | |
| "learning_rate": 0.0001684422110552764, | |
| "loss": 2.597, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.4616467276565799, | |
| "grad_norm": 0.5229634046554565, | |
| "learning_rate": 0.00016824120603015078, | |
| "loss": 2.723, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.4644616467276566, | |
| "grad_norm": 0.6184930801391602, | |
| "learning_rate": 0.00016804020100502512, | |
| "loss": 2.703, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.4672765657987333, | |
| "grad_norm": 0.6252800226211548, | |
| "learning_rate": 0.0001678391959798995, | |
| "loss": 2.5517, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.47009148486981, | |
| "grad_norm": 0.5429969429969788, | |
| "learning_rate": 0.0001676381909547739, | |
| "loss": 2.7248, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.4729064039408867, | |
| "grad_norm": 0.6234527230262756, | |
| "learning_rate": 0.00016743718592964827, | |
| "loss": 2.5844, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.4757213230119634, | |
| "grad_norm": 0.6872987747192383, | |
| "learning_rate": 0.0001672361809045226, | |
| "loss": 2.5067, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.47853624208304013, | |
| "grad_norm": 0.5591785907745361, | |
| "learning_rate": 0.00016703517587939699, | |
| "loss": 2.4999, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.4813511611541168, | |
| "grad_norm": 0.5767291188240051, | |
| "learning_rate": 0.00016683417085427136, | |
| "loss": 2.8885, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.48416608022519353, | |
| "grad_norm": 0.6422219276428223, | |
| "learning_rate": 0.00016663316582914573, | |
| "loss": 2.6989, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.48698099929627026, | |
| "grad_norm": 0.6035985350608826, | |
| "learning_rate": 0.0001664321608040201, | |
| "loss": 2.6412, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.4897959183673469, | |
| "grad_norm": 0.5744962096214294, | |
| "learning_rate": 0.00016623115577889448, | |
| "loss": 2.6395, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.49261083743842365, | |
| "grad_norm": 0.6550725698471069, | |
| "learning_rate": 0.00016603015075376885, | |
| "loss": 2.7526, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4954257565095003, | |
| "grad_norm": 0.7883411049842834, | |
| "learning_rate": 0.00016582914572864322, | |
| "loss": 2.5775, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.49824067558057705, | |
| "grad_norm": 0.6014293432235718, | |
| "learning_rate": 0.0001656281407035176, | |
| "loss": 2.6568, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.5010555946516537, | |
| "grad_norm": 0.5285369157791138, | |
| "learning_rate": 0.00016542713567839197, | |
| "loss": 2.4862, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.5038705137227305, | |
| "grad_norm": 0.6501176953315735, | |
| "learning_rate": 0.00016522613065326634, | |
| "loss": 2.631, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.5066854327938072, | |
| "grad_norm": 0.6903632879257202, | |
| "learning_rate": 0.00016502512562814072, | |
| "loss": 2.7363, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.5095003518648838, | |
| "grad_norm": 0.6202127933502197, | |
| "learning_rate": 0.0001648241206030151, | |
| "loss": 2.6678, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.5123152709359606, | |
| "grad_norm": 0.688332200050354, | |
| "learning_rate": 0.00016462311557788946, | |
| "loss": 2.6346, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.5151301900070373, | |
| "grad_norm": 0.5428361892700195, | |
| "learning_rate": 0.0001644221105527638, | |
| "loss": 2.9607, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.517945109078114, | |
| "grad_norm": 0.5395454168319702, | |
| "learning_rate": 0.0001642211055276382, | |
| "loss": 2.8427, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5207600281491908, | |
| "grad_norm": 0.554793655872345, | |
| "learning_rate": 0.00016402010050251258, | |
| "loss": 2.4523, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5235749472202674, | |
| "grad_norm": 0.5698427557945251, | |
| "learning_rate": 0.00016381909547738695, | |
| "loss": 2.7052, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.5263898662913441, | |
| "grad_norm": 0.5099812150001526, | |
| "learning_rate": 0.0001636180904522613, | |
| "loss": 2.3651, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.5292047853624209, | |
| "grad_norm": 0.5726649761199951, | |
| "learning_rate": 0.0001634170854271357, | |
| "loss": 2.9888, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5320197044334976, | |
| "grad_norm": 0.630757212638855, | |
| "learning_rate": 0.00016321608040201007, | |
| "loss": 2.6886, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.5348346235045742, | |
| "grad_norm": 0.5425901412963867, | |
| "learning_rate": 0.00016301507537688442, | |
| "loss": 2.6047, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5376495425756509, | |
| "grad_norm": 0.6030427813529968, | |
| "learning_rate": 0.0001628140703517588, | |
| "loss": 2.8713, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.5404644616467277, | |
| "grad_norm": 0.5966265201568604, | |
| "learning_rate": 0.00016261306532663316, | |
| "loss": 2.7212, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.5432793807178043, | |
| "grad_norm": 0.5909569263458252, | |
| "learning_rate": 0.00016241206030150756, | |
| "loss": 2.8696, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.546094299788881, | |
| "grad_norm": 0.5485597252845764, | |
| "learning_rate": 0.0001622110552763819, | |
| "loss": 2.6367, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.5489092188599578, | |
| "grad_norm": 0.6274406909942627, | |
| "learning_rate": 0.00016201005025125628, | |
| "loss": 2.6882, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.5517241379310345, | |
| "grad_norm": 0.5513599514961243, | |
| "learning_rate": 0.00016180904522613066, | |
| "loss": 2.6596, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.5545390570021111, | |
| "grad_norm": 0.7499818801879883, | |
| "learning_rate": 0.00016160804020100503, | |
| "loss": 2.6191, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.5573539760731879, | |
| "grad_norm": 0.5143455862998962, | |
| "learning_rate": 0.0001614070351758794, | |
| "loss": 2.6543, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.5601688951442646, | |
| "grad_norm": 0.6372074484825134, | |
| "learning_rate": 0.00016120603015075378, | |
| "loss": 2.4355, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.5629838142153413, | |
| "grad_norm": 0.7132628560066223, | |
| "learning_rate": 0.00016100502512562815, | |
| "loss": 2.671, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.565798733286418, | |
| "grad_norm": 0.5970779657363892, | |
| "learning_rate": 0.00016080402010050252, | |
| "loss": 2.6802, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.5686136523574947, | |
| "grad_norm": 0.6065824627876282, | |
| "learning_rate": 0.0001606030150753769, | |
| "loss": 2.6151, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 0.5521674752235413, | |
| "learning_rate": 0.00016040201005025127, | |
| "loss": 2.5132, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.5742434904996482, | |
| "grad_norm": 0.6067800521850586, | |
| "learning_rate": 0.00016020100502512564, | |
| "loss": 2.9283, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.5770584095707249, | |
| "grad_norm": 0.5979752540588379, | |
| "learning_rate": 0.00016, | |
| "loss": 2.5355, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.5798733286418015, | |
| "grad_norm": 0.6044461727142334, | |
| "learning_rate": 0.00015979899497487439, | |
| "loss": 2.6752, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.5826882477128783, | |
| "grad_norm": 0.580636739730835, | |
| "learning_rate": 0.00015959798994974876, | |
| "loss": 2.538, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.585503166783955, | |
| "grad_norm": 0.6181825995445251, | |
| "learning_rate": 0.0001593969849246231, | |
| "loss": 2.6529, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.5883180858550316, | |
| "grad_norm": 0.6641463041305542, | |
| "learning_rate": 0.0001591959798994975, | |
| "loss": 2.6252, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.5911330049261084, | |
| "grad_norm": 0.6233858466148376, | |
| "learning_rate": 0.00015899497487437188, | |
| "loss": 2.6832, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.5939479239971851, | |
| "grad_norm": 0.6696732044219971, | |
| "learning_rate": 0.00015879396984924625, | |
| "loss": 3.0325, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.5967628430682618, | |
| "grad_norm": 0.7569646239280701, | |
| "learning_rate": 0.0001585929648241206, | |
| "loss": 2.675, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.5995777621393384, | |
| "grad_norm": 0.5962279438972473, | |
| "learning_rate": 0.000158391959798995, | |
| "loss": 2.6369, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.6023926812104152, | |
| "grad_norm": 0.6349969506263733, | |
| "learning_rate": 0.00015819095477386937, | |
| "loss": 2.5531, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.6052076002814919, | |
| "grad_norm": 0.8234291076660156, | |
| "learning_rate": 0.00015798994974874372, | |
| "loss": 2.5466, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.6080225193525686, | |
| "grad_norm": 0.6057316660881042, | |
| "learning_rate": 0.0001577889447236181, | |
| "loss": 2.3296, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.6108374384236454, | |
| "grad_norm": 0.6568176746368408, | |
| "learning_rate": 0.00015758793969849246, | |
| "loss": 2.8075, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.613652357494722, | |
| "grad_norm": 0.5945923328399658, | |
| "learning_rate": 0.00015738693467336686, | |
| "loss": 2.696, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.6164672765657987, | |
| "grad_norm": 0.6226676106452942, | |
| "learning_rate": 0.0001571859296482412, | |
| "loss": 2.5764, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.6192821956368755, | |
| "grad_norm": 0.6158185601234436, | |
| "learning_rate": 0.00015698492462311558, | |
| "loss": 2.66, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6220971147079521, | |
| "grad_norm": 0.7033487558364868, | |
| "learning_rate": 0.00015678391959798995, | |
| "loss": 2.7747, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.6249120337790288, | |
| "grad_norm": 0.5215992331504822, | |
| "learning_rate": 0.00015658291457286433, | |
| "loss": 2.4176, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.6277269528501056, | |
| "grad_norm": 0.8559087514877319, | |
| "learning_rate": 0.0001563819095477387, | |
| "loss": 2.8081, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.6305418719211823, | |
| "grad_norm": 0.5106130242347717, | |
| "learning_rate": 0.00015618090452261307, | |
| "loss": 2.6433, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.633356790992259, | |
| "grad_norm": 0.6176455020904541, | |
| "learning_rate": 0.00015597989949748745, | |
| "loss": 2.4351, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.6361717100633357, | |
| "grad_norm": 0.8193095922470093, | |
| "learning_rate": 0.00015577889447236182, | |
| "loss": 2.8882, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.6389866291344124, | |
| "grad_norm": 0.8569721579551697, | |
| "learning_rate": 0.0001555778894472362, | |
| "loss": 2.7263, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.6418015482054891, | |
| "grad_norm": 0.6688103079795837, | |
| "learning_rate": 0.00015537688442211056, | |
| "loss": 2.6202, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.6446164672765659, | |
| "grad_norm": 0.6070395708084106, | |
| "learning_rate": 0.00015517587939698494, | |
| "loss": 2.7117, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.6474313863476425, | |
| "grad_norm": 0.7812969088554382, | |
| "learning_rate": 0.0001549748743718593, | |
| "loss": 2.5815, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6502463054187192, | |
| "grad_norm": 0.8165440559387207, | |
| "learning_rate": 0.00015477386934673368, | |
| "loss": 2.8503, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.6530612244897959, | |
| "grad_norm": 0.6454505324363708, | |
| "learning_rate": 0.00015457286432160806, | |
| "loss": 2.5918, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.6558761435608726, | |
| "grad_norm": 0.7109069228172302, | |
| "learning_rate": 0.0001543718592964824, | |
| "loss": 2.5554, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.6586910626319493, | |
| "grad_norm": 0.6079565286636353, | |
| "learning_rate": 0.0001541708542713568, | |
| "loss": 2.6232, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.661505981703026, | |
| "grad_norm": 0.576082170009613, | |
| "learning_rate": 0.00015396984924623117, | |
| "loss": 2.8387, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.6643209007741028, | |
| "grad_norm": 0.5683891177177429, | |
| "learning_rate": 0.00015376884422110555, | |
| "loss": 2.7391, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.6671358198451794, | |
| "grad_norm": 0.6114887595176697, | |
| "learning_rate": 0.0001535678391959799, | |
| "loss": 2.6629, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.6699507389162561, | |
| "grad_norm": 0.6666116118431091, | |
| "learning_rate": 0.00015336683417085427, | |
| "loss": 2.9027, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.6727656579873329, | |
| "grad_norm": 0.5646522641181946, | |
| "learning_rate": 0.00015316582914572867, | |
| "loss": 2.7252, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.6755805770584096, | |
| "grad_norm": 0.6885817646980286, | |
| "learning_rate": 0.000152964824120603, | |
| "loss": 2.5966, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.6783954961294862, | |
| "grad_norm": 0.5778309106826782, | |
| "learning_rate": 0.00015276381909547739, | |
| "loss": 2.6275, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.681210415200563, | |
| "grad_norm": 0.6230787038803101, | |
| "learning_rate": 0.00015256281407035176, | |
| "loss": 2.5258, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.6840253342716397, | |
| "grad_norm": 0.5411630272865295, | |
| "learning_rate": 0.00015236180904522613, | |
| "loss": 2.6047, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.6868402533427164, | |
| "grad_norm": 0.5547896027565002, | |
| "learning_rate": 0.0001521608040201005, | |
| "loss": 2.6451, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.6896551724137931, | |
| "grad_norm": 0.6721991300582886, | |
| "learning_rate": 0.00015195979899497488, | |
| "loss": 2.7127, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.6924700914848698, | |
| "grad_norm": 0.6402304172515869, | |
| "learning_rate": 0.00015175879396984925, | |
| "loss": 2.8568, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.6952850105559465, | |
| "grad_norm": 0.594251275062561, | |
| "learning_rate": 0.00015155778894472362, | |
| "loss": 2.7168, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.6980999296270233, | |
| "grad_norm": 0.8853170871734619, | |
| "learning_rate": 0.000151356783919598, | |
| "loss": 2.7259, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.7009148486981, | |
| "grad_norm": 0.5619581341743469, | |
| "learning_rate": 0.00015115577889447237, | |
| "loss": 2.6218, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.7037297677691766, | |
| "grad_norm": 0.6149075031280518, | |
| "learning_rate": 0.00015095477386934674, | |
| "loss": 2.8984, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.7065446868402533, | |
| "grad_norm": 0.6819274425506592, | |
| "learning_rate": 0.00015075376884422112, | |
| "loss": 2.8293, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.7093596059113301, | |
| "grad_norm": 0.5911348462104797, | |
| "learning_rate": 0.0001505527638190955, | |
| "loss": 2.6333, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.7121745249824067, | |
| "grad_norm": 0.7064481973648071, | |
| "learning_rate": 0.00015035175879396986, | |
| "loss": 2.8061, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.7149894440534834, | |
| "grad_norm": 0.6039316654205322, | |
| "learning_rate": 0.00015015075376884423, | |
| "loss": 2.6468, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.7178043631245602, | |
| "grad_norm": 0.5624644756317139, | |
| "learning_rate": 0.0001499497487437186, | |
| "loss": 2.7813, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.7206192821956369, | |
| "grad_norm": 0.5971612334251404, | |
| "learning_rate": 0.00014974874371859298, | |
| "loss": 2.6411, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.7234342012667135, | |
| "grad_norm": 0.6717031598091125, | |
| "learning_rate": 0.00014954773869346735, | |
| "loss": 2.5579, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.7262491203377903, | |
| "grad_norm": 0.6643320322036743, | |
| "learning_rate": 0.0001493467336683417, | |
| "loss": 2.8469, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.729064039408867, | |
| "grad_norm": 0.5971053838729858, | |
| "learning_rate": 0.0001491457286432161, | |
| "loss": 2.613, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.7318789584799437, | |
| "grad_norm": 0.6267710328102112, | |
| "learning_rate": 0.00014894472361809047, | |
| "loss": 2.6969, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.7346938775510204, | |
| "grad_norm": 0.6237425804138184, | |
| "learning_rate": 0.00014874371859296482, | |
| "loss": 2.7047, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.7375087966220971, | |
| "grad_norm": 0.5603229999542236, | |
| "learning_rate": 0.0001485427135678392, | |
| "loss": 2.7534, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.7403237156931738, | |
| "grad_norm": 0.6484439969062805, | |
| "learning_rate": 0.00014834170854271356, | |
| "loss": 2.9389, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.7431386347642506, | |
| "grad_norm": 0.6225891709327698, | |
| "learning_rate": 0.00014814070351758796, | |
| "loss": 2.7516, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.7459535538353272, | |
| "grad_norm": 0.5303828716278076, | |
| "learning_rate": 0.0001479396984924623, | |
| "loss": 2.3335, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.7487684729064039, | |
| "grad_norm": 0.6280227303504944, | |
| "learning_rate": 0.00014773869346733668, | |
| "loss": 2.6259, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.7515833919774807, | |
| "grad_norm": 0.5551609992980957, | |
| "learning_rate": 0.00014753768844221106, | |
| "loss": 2.6572, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.7543983110485574, | |
| "grad_norm": 0.7833865284919739, | |
| "learning_rate": 0.00014733668341708543, | |
| "loss": 2.7358, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.757213230119634, | |
| "grad_norm": 0.6138265132904053, | |
| "learning_rate": 0.0001471356783919598, | |
| "loss": 2.4282, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.7600281491907108, | |
| "grad_norm": 0.6331743001937866, | |
| "learning_rate": 0.00014693467336683417, | |
| "loss": 2.6373, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.7628430682617875, | |
| "grad_norm": 0.569272518157959, | |
| "learning_rate": 0.00014673366834170855, | |
| "loss": 2.6546, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.7656579873328642, | |
| "grad_norm": 0.6755379438400269, | |
| "learning_rate": 0.00014653266331658292, | |
| "loss": 2.6258, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.7684729064039408, | |
| "grad_norm": 0.6408460140228271, | |
| "learning_rate": 0.0001463316582914573, | |
| "loss": 2.6824, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.7712878254750176, | |
| "grad_norm": 0.6325194239616394, | |
| "learning_rate": 0.00014613065326633167, | |
| "loss": 2.7959, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.7741027445460943, | |
| "grad_norm": 0.6526459455490112, | |
| "learning_rate": 0.00014592964824120604, | |
| "loss": 2.613, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.776917663617171, | |
| "grad_norm": 0.610998272895813, | |
| "learning_rate": 0.0001457286432160804, | |
| "loss": 2.4973, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.7797325826882477, | |
| "grad_norm": 0.510045051574707, | |
| "learning_rate": 0.00014552763819095479, | |
| "loss": 2.7343, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.7825475017593244, | |
| "grad_norm": 0.5863422155380249, | |
| "learning_rate": 0.00014532663316582916, | |
| "loss": 2.5543, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.7853624208304011, | |
| "grad_norm": 0.5406447649002075, | |
| "learning_rate": 0.00014512562814070353, | |
| "loss": 2.7748, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.7881773399014779, | |
| "grad_norm": 0.7465657591819763, | |
| "learning_rate": 0.0001449246231155779, | |
| "loss": 2.4034, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.7909922589725545, | |
| "grad_norm": 0.5192904472351074, | |
| "learning_rate": 0.00014472361809045228, | |
| "loss": 2.4881, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.7938071780436312, | |
| "grad_norm": 0.6085344552993774, | |
| "learning_rate": 0.00014452261306532665, | |
| "loss": 2.6534, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.796622097114708, | |
| "grad_norm": 0.6155668497085571, | |
| "learning_rate": 0.000144321608040201, | |
| "loss": 2.6149, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.7994370161857847, | |
| "grad_norm": 0.623285710811615, | |
| "learning_rate": 0.00014412060301507537, | |
| "loss": 2.7725, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.8022519352568613, | |
| "grad_norm": 0.9461747407913208, | |
| "learning_rate": 0.00014391959798994977, | |
| "loss": 2.7837, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.8050668543279381, | |
| "grad_norm": 0.7152134776115417, | |
| "learning_rate": 0.00014371859296482411, | |
| "loss": 2.6678, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.8078817733990148, | |
| "grad_norm": 0.6519983410835266, | |
| "learning_rate": 0.0001435175879396985, | |
| "loss": 2.5234, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.8106966924700915, | |
| "grad_norm": 0.6523590087890625, | |
| "learning_rate": 0.00014331658291457286, | |
| "loss": 2.7284, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.8135116115411682, | |
| "grad_norm": 0.6067202687263489, | |
| "learning_rate": 0.00014311557788944726, | |
| "loss": 2.5459, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 0.5836743116378784, | |
| "learning_rate": 0.0001429145728643216, | |
| "loss": 2.452, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.8191414496833216, | |
| "grad_norm": 0.685727596282959, | |
| "learning_rate": 0.00014271356783919598, | |
| "loss": 2.6568, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.8219563687543983, | |
| "grad_norm": 0.6456769704818726, | |
| "learning_rate": 0.00014251256281407035, | |
| "loss": 2.6553, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.824771287825475, | |
| "grad_norm": 0.6357674598693848, | |
| "learning_rate": 0.00014231155778894473, | |
| "loss": 2.7589, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.8275862068965517, | |
| "grad_norm": 0.6339374780654907, | |
| "learning_rate": 0.0001421105527638191, | |
| "loss": 2.7797, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.8304011259676284, | |
| "grad_norm": 0.5491819381713867, | |
| "learning_rate": 0.00014190954773869347, | |
| "loss": 2.4772, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.8332160450387052, | |
| "grad_norm": 0.6312305331230164, | |
| "learning_rate": 0.00014170854271356784, | |
| "loss": 2.6715, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.8360309641097818, | |
| "grad_norm": 0.647985577583313, | |
| "learning_rate": 0.00014150753768844222, | |
| "loss": 2.7811, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.8388458831808585, | |
| "grad_norm": 0.6383928060531616, | |
| "learning_rate": 0.0001413065326633166, | |
| "loss": 2.8496, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.8416608022519353, | |
| "grad_norm": 0.5548710823059082, | |
| "learning_rate": 0.00014110552763819096, | |
| "loss": 2.5096, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.844475721323012, | |
| "grad_norm": 0.5331722497940063, | |
| "learning_rate": 0.00014090452261306534, | |
| "loss": 2.5683, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8472906403940886, | |
| "grad_norm": 0.5956087112426758, | |
| "learning_rate": 0.0001407035175879397, | |
| "loss": 2.6175, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.8501055594651654, | |
| "grad_norm": 0.6151571273803711, | |
| "learning_rate": 0.00014050251256281408, | |
| "loss": 2.5863, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.8529204785362421, | |
| "grad_norm": 0.5952453017234802, | |
| "learning_rate": 0.00014030150753768846, | |
| "loss": 2.5152, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.8557353976073188, | |
| "grad_norm": 0.6127233505249023, | |
| "learning_rate": 0.0001401005025125628, | |
| "loss": 2.5753, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.8585503166783955, | |
| "grad_norm": 0.551474928855896, | |
| "learning_rate": 0.0001398994974874372, | |
| "loss": 2.6523, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.8613652357494722, | |
| "grad_norm": 0.8345268368721008, | |
| "learning_rate": 0.00013969849246231157, | |
| "loss": 2.78, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.8641801548205489, | |
| "grad_norm": 0.6494585275650024, | |
| "learning_rate": 0.00013949748743718595, | |
| "loss": 2.7354, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.8669950738916257, | |
| "grad_norm": 0.6813188791275024, | |
| "learning_rate": 0.0001392964824120603, | |
| "loss": 2.8736, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.8698099929627023, | |
| "grad_norm": 0.6250954270362854, | |
| "learning_rate": 0.00013909547738693467, | |
| "loss": 2.7548, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.872624912033779, | |
| "grad_norm": 0.6115372180938721, | |
| "learning_rate": 0.00013889447236180907, | |
| "loss": 2.7126, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.8754398311048557, | |
| "grad_norm": 0.6005333662033081, | |
| "learning_rate": 0.0001386934673366834, | |
| "loss": 2.9623, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.8782547501759325, | |
| "grad_norm": 0.5203389525413513, | |
| "learning_rate": 0.00013849246231155778, | |
| "loss": 2.7269, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.8810696692470091, | |
| "grad_norm": 0.5951765775680542, | |
| "learning_rate": 0.00013829145728643216, | |
| "loss": 2.6058, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.8838845883180858, | |
| "grad_norm": 0.6142780184745789, | |
| "learning_rate": 0.00013809045226130656, | |
| "loss": 2.5918, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.8866995073891626, | |
| "grad_norm": 0.5776972770690918, | |
| "learning_rate": 0.0001378894472361809, | |
| "loss": 2.5006, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.8895144264602393, | |
| "grad_norm": 0.6553467512130737, | |
| "learning_rate": 0.00013768844221105528, | |
| "loss": 2.5903, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.8923293455313159, | |
| "grad_norm": 0.5776195526123047, | |
| "learning_rate": 0.00013748743718592965, | |
| "loss": 2.5532, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.8951442646023927, | |
| "grad_norm": 0.5531054139137268, | |
| "learning_rate": 0.00013728643216080402, | |
| "loss": 2.624, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.8979591836734694, | |
| "grad_norm": 0.7402701377868652, | |
| "learning_rate": 0.0001370854271356784, | |
| "loss": 2.7686, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.9007741027445461, | |
| "grad_norm": 0.5394028425216675, | |
| "learning_rate": 0.00013688442211055277, | |
| "loss": 2.6044, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.9035890218156228, | |
| "grad_norm": 0.6454526782035828, | |
| "learning_rate": 0.00013668341708542714, | |
| "loss": 2.6938, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.9064039408866995, | |
| "grad_norm": 0.7545249462127686, | |
| "learning_rate": 0.00013648241206030151, | |
| "loss": 2.5263, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.9092188599577762, | |
| "grad_norm": 0.6479030251502991, | |
| "learning_rate": 0.0001362814070351759, | |
| "loss": 2.4625, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.912033779028853, | |
| "grad_norm": 0.9134926199913025, | |
| "learning_rate": 0.00013608040201005026, | |
| "loss": 2.7668, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.9148486980999296, | |
| "grad_norm": 0.6736027002334595, | |
| "learning_rate": 0.00013587939698492463, | |
| "loss": 2.826, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.9176636171710063, | |
| "grad_norm": 0.6161238551139832, | |
| "learning_rate": 0.000135678391959799, | |
| "loss": 2.513, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.9204785362420831, | |
| "grad_norm": 0.7301089763641357, | |
| "learning_rate": 0.00013547738693467338, | |
| "loss": 2.6374, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.9232934553131598, | |
| "grad_norm": 0.5782633423805237, | |
| "learning_rate": 0.00013527638190954775, | |
| "loss": 2.4847, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.9261083743842364, | |
| "grad_norm": 0.6025380492210388, | |
| "learning_rate": 0.0001350753768844221, | |
| "loss": 2.6272, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.9289232934553132, | |
| "grad_norm": 0.6242662668228149, | |
| "learning_rate": 0.00013487437185929647, | |
| "loss": 2.801, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.9317382125263899, | |
| "grad_norm": 0.7133350372314453, | |
| "learning_rate": 0.00013467336683417087, | |
| "loss": 2.5261, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.9345531315974666, | |
| "grad_norm": 0.5895963311195374, | |
| "learning_rate": 0.00013447236180904524, | |
| "loss": 2.7813, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.9373680506685432, | |
| "grad_norm": 0.7254224419593811, | |
| "learning_rate": 0.0001342713567839196, | |
| "loss": 2.6859, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.94018296973962, | |
| "grad_norm": 0.7255984544754028, | |
| "learning_rate": 0.00013407035175879396, | |
| "loss": 2.6328, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.9429978888106967, | |
| "grad_norm": 0.827979564666748, | |
| "learning_rate": 0.00013386934673366836, | |
| "loss": 2.7311, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.9458128078817734, | |
| "grad_norm": 0.6603137850761414, | |
| "learning_rate": 0.0001336683417085427, | |
| "loss": 2.6389, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.9486277269528501, | |
| "grad_norm": 0.6362401247024536, | |
| "learning_rate": 0.00013346733668341708, | |
| "loss": 2.7471, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.9514426460239268, | |
| "grad_norm": 0.7190608382225037, | |
| "learning_rate": 0.00013326633165829146, | |
| "loss": 2.5964, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.9542575650950035, | |
| "grad_norm": 0.6639814376831055, | |
| "learning_rate": 0.00013306532663316586, | |
| "loss": 2.6358, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.9570724841660803, | |
| "grad_norm": 0.6791893243789673, | |
| "learning_rate": 0.0001328643216080402, | |
| "loss": 2.6764, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.9598874032371569, | |
| "grad_norm": 0.6663180589675903, | |
| "learning_rate": 0.00013266331658291457, | |
| "loss": 2.8057, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.9627023223082336, | |
| "grad_norm": 0.5866056680679321, | |
| "learning_rate": 0.00013246231155778895, | |
| "loss": 2.6845, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.9655172413793104, | |
| "grad_norm": 0.6590510010719299, | |
| "learning_rate": 0.00013226130653266332, | |
| "loss": 2.4007, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.9683321604503871, | |
| "grad_norm": 0.6014566421508789, | |
| "learning_rate": 0.0001320603015075377, | |
| "loss": 2.4501, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.9711470795214637, | |
| "grad_norm": 0.7037169337272644, | |
| "learning_rate": 0.00013185929648241207, | |
| "loss": 3.0925, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.9739619985925405, | |
| "grad_norm": 0.5314791798591614, | |
| "learning_rate": 0.00013165829145728644, | |
| "loss": 2.4216, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.9767769176636172, | |
| "grad_norm": 0.5568397045135498, | |
| "learning_rate": 0.0001314572864321608, | |
| "loss": 2.4926, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.9795918367346939, | |
| "grad_norm": 0.728171706199646, | |
| "learning_rate": 0.00013125628140703518, | |
| "loss": 2.6955, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.9824067558057706, | |
| "grad_norm": 0.7184565663337708, | |
| "learning_rate": 0.00013105527638190956, | |
| "loss": 2.6989, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.9852216748768473, | |
| "grad_norm": 1.116127610206604, | |
| "learning_rate": 0.00013085427135678393, | |
| "loss": 2.5996, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.988036593947924, | |
| "grad_norm": 0.6320902705192566, | |
| "learning_rate": 0.0001306532663316583, | |
| "loss": 2.571, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.9908515130190007, | |
| "grad_norm": 0.6517199277877808, | |
| "learning_rate": 0.00013045226130653268, | |
| "loss": 2.6731, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.9936664320900774, | |
| "grad_norm": 0.6911283135414124, | |
| "learning_rate": 0.00013025125628140705, | |
| "loss": 2.8807, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.9964813511611541, | |
| "grad_norm": 0.7019006609916687, | |
| "learning_rate": 0.0001300502512562814, | |
| "loss": 2.6346, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.9992962702322308, | |
| "grad_norm": 0.6711909770965576, | |
| "learning_rate": 0.00012984924623115577, | |
| "loss": 2.7176, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.0021111893033074, | |
| "grad_norm": 0.6791936755180359, | |
| "learning_rate": 0.00012964824120603017, | |
| "loss": 2.4828, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.0049261083743843, | |
| "grad_norm": 0.550987720489502, | |
| "learning_rate": 0.00012944723618090454, | |
| "loss": 2.4396, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.007741027445461, | |
| "grad_norm": 0.6731054186820984, | |
| "learning_rate": 0.0001292462311557789, | |
| "loss": 2.7791, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.0105559465165377, | |
| "grad_norm": 0.5614567995071411, | |
| "learning_rate": 0.00012904522613065326, | |
| "loss": 2.5572, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.0133708655876144, | |
| "grad_norm": 0.5224441289901733, | |
| "learning_rate": 0.00012884422110552766, | |
| "loss": 2.3983, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.016185784658691, | |
| "grad_norm": 0.534264326095581, | |
| "learning_rate": 0.000128643216080402, | |
| "loss": 2.4682, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.0190007037297677, | |
| "grad_norm": 0.7560765743255615, | |
| "learning_rate": 0.00012844221105527638, | |
| "loss": 2.5593, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.0218156228008444, | |
| "grad_norm": 0.6609757542610168, | |
| "learning_rate": 0.00012824120603015075, | |
| "loss": 2.4428, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.0246305418719213, | |
| "grad_norm": 0.48449280858039856, | |
| "learning_rate": 0.00012804020100502515, | |
| "loss": 2.2216, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.027445460942998, | |
| "grad_norm": 0.6201764345169067, | |
| "learning_rate": 0.0001278391959798995, | |
| "loss": 2.4076, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.0302603800140746, | |
| "grad_norm": 0.6022098660469055, | |
| "learning_rate": 0.00012763819095477387, | |
| "loss": 2.3617, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.0330752990851513, | |
| "grad_norm": 0.5485665798187256, | |
| "learning_rate": 0.00012743718592964824, | |
| "loss": 2.4646, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.035890218156228, | |
| "grad_norm": 0.6300007700920105, | |
| "learning_rate": 0.00012723618090452262, | |
| "loss": 2.3845, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.0387051372273046, | |
| "grad_norm": 0.6588097214698792, | |
| "learning_rate": 0.000127035175879397, | |
| "loss": 2.3836, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.0415200562983815, | |
| "grad_norm": 0.6159886717796326, | |
| "learning_rate": 0.00012683417085427136, | |
| "loss": 2.239, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.0443349753694582, | |
| "grad_norm": 0.7142757177352905, | |
| "learning_rate": 0.00012663316582914574, | |
| "loss": 2.3208, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.0471498944405349, | |
| "grad_norm": 0.7620591521263123, | |
| "learning_rate": 0.0001264321608040201, | |
| "loss": 2.5665, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.0499648135116115, | |
| "grad_norm": 0.6486737728118896, | |
| "learning_rate": 0.00012623115577889448, | |
| "loss": 2.8276, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.0527797325826882, | |
| "grad_norm": 0.622787594795227, | |
| "learning_rate": 0.00012603015075376885, | |
| "loss": 2.4826, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.0555946516537649, | |
| "grad_norm": 0.6556206345558167, | |
| "learning_rate": 0.00012582914572864323, | |
| "loss": 2.5765, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.0584095707248418, | |
| "grad_norm": 0.6902799606323242, | |
| "learning_rate": 0.0001256281407035176, | |
| "loss": 2.3851, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.0612244897959184, | |
| "grad_norm": 0.6362977027893066, | |
| "learning_rate": 0.00012542713567839197, | |
| "loss": 2.4587, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.064039408866995, | |
| "grad_norm": 0.6027363538742065, | |
| "learning_rate": 0.00012522613065326635, | |
| "loss": 2.3945, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.0668543279380718, | |
| "grad_norm": 0.681010365486145, | |
| "learning_rate": 0.0001250251256281407, | |
| "loss": 2.247, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.0696692470091484, | |
| "grad_norm": 0.588394284248352, | |
| "learning_rate": 0.00012482412060301507, | |
| "loss": 2.3279, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.0724841660802251, | |
| "grad_norm": 0.6285263299942017, | |
| "learning_rate": 0.00012462311557788947, | |
| "loss": 2.2176, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.0752990851513018, | |
| "grad_norm": 0.6699137091636658, | |
| "learning_rate": 0.00012442211055276384, | |
| "loss": 2.4252, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.0781140042223787, | |
| "grad_norm": 0.7217219471931458, | |
| "learning_rate": 0.00012422110552763818, | |
| "loss": 2.3409, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.0809289232934554, | |
| "grad_norm": 0.6710893511772156, | |
| "learning_rate": 0.00012402010050251256, | |
| "loss": 2.4054, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.083743842364532, | |
| "grad_norm": 0.665313720703125, | |
| "learning_rate": 0.00012381909547738696, | |
| "loss": 2.4749, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.0865587614356087, | |
| "grad_norm": 0.655486524105072, | |
| "learning_rate": 0.0001236180904522613, | |
| "loss": 2.1261, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.0893736805066854, | |
| "grad_norm": 0.9402002692222595, | |
| "learning_rate": 0.00012341708542713568, | |
| "loss": 2.5344, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.092188599577762, | |
| "grad_norm": 0.760830819606781, | |
| "learning_rate": 0.00012321608040201005, | |
| "loss": 2.6215, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.095003518648839, | |
| "grad_norm": 0.7938470244407654, | |
| "learning_rate": 0.00012301507537688445, | |
| "loss": 2.3226, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.0978184377199156, | |
| "grad_norm": 0.789606511592865, | |
| "learning_rate": 0.0001228140703517588, | |
| "loss": 2.4419, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.1006333567909923, | |
| "grad_norm": 0.7316797375679016, | |
| "learning_rate": 0.00012261306532663317, | |
| "loss": 2.6489, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.103448275862069, | |
| "grad_norm": 0.7367439270019531, | |
| "learning_rate": 0.00012241206030150754, | |
| "loss": 2.2951, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.1062631949331456, | |
| "grad_norm": 0.735031008720398, | |
| "learning_rate": 0.00012221105527638191, | |
| "loss": 2.3752, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.1090781140042223, | |
| "grad_norm": 0.8442686796188354, | |
| "learning_rate": 0.00012201005025125629, | |
| "loss": 2.4054, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.1118930330752992, | |
| "grad_norm": 0.7112425565719604, | |
| "learning_rate": 0.00012180904522613066, | |
| "loss": 2.4153, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.1147079521463759, | |
| "grad_norm": 0.8225473165512085, | |
| "learning_rate": 0.00012160804020100502, | |
| "loss": 2.3042, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.1175228712174525, | |
| "grad_norm": 0.8238793015480042, | |
| "learning_rate": 0.00012140703517587942, | |
| "loss": 2.3454, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.1203377902885292, | |
| "grad_norm": 0.9430282711982727, | |
| "learning_rate": 0.00012120603015075378, | |
| "loss": 2.4815, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.1231527093596059, | |
| "grad_norm": 0.8851016759872437, | |
| "learning_rate": 0.00012100502512562815, | |
| "loss": 2.2304, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.1259676284306825, | |
| "grad_norm": 1.1577056646347046, | |
| "learning_rate": 0.00012080402010050251, | |
| "loss": 2.3881, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.1287825475017592, | |
| "grad_norm": 0.8923066258430481, | |
| "learning_rate": 0.00012060301507537688, | |
| "loss": 2.6341, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.131597466572836, | |
| "grad_norm": 0.8544619679450989, | |
| "learning_rate": 0.00012040201005025127, | |
| "loss": 2.5802, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.1344123856439128, | |
| "grad_norm": 0.9601594805717468, | |
| "learning_rate": 0.00012020100502512563, | |
| "loss": 2.2292, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.1372273047149895, | |
| "grad_norm": 0.9403390884399414, | |
| "learning_rate": 0.00012, | |
| "loss": 2.5492, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.1400422237860661, | |
| "grad_norm": 0.7530049681663513, | |
| "learning_rate": 0.00011979899497487436, | |
| "loss": 2.3684, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.1428571428571428, | |
| "grad_norm": 1.0320841073989868, | |
| "learning_rate": 0.00011959798994974876, | |
| "loss": 2.4157, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.1456720619282197, | |
| "grad_norm": 1.1246405839920044, | |
| "learning_rate": 0.00011939698492462312, | |
| "loss": 2.6815, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.1484869809992964, | |
| "grad_norm": 0.8500766158103943, | |
| "learning_rate": 0.0001191959798994975, | |
| "loss": 2.4088, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.151301900070373, | |
| "grad_norm": 0.7606078386306763, | |
| "learning_rate": 0.00011899497487437185, | |
| "loss": 2.2564, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.1541168191414497, | |
| "grad_norm": 0.8513486385345459, | |
| "learning_rate": 0.00011879396984924624, | |
| "loss": 2.5881, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.1569317382125264, | |
| "grad_norm": 0.7827906012535095, | |
| "learning_rate": 0.00011859296482412061, | |
| "loss": 2.4386, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.159746657283603, | |
| "grad_norm": 0.9784967303276062, | |
| "learning_rate": 0.00011839195979899497, | |
| "loss": 2.5552, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.1625615763546797, | |
| "grad_norm": 0.8472895622253418, | |
| "learning_rate": 0.00011819095477386935, | |
| "loss": 2.6518, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.1653764954257566, | |
| "grad_norm": 0.7687847018241882, | |
| "learning_rate": 0.00011798994974874373, | |
| "loss": 2.39, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.1681914144968333, | |
| "grad_norm": 0.7497126460075378, | |
| "learning_rate": 0.0001177889447236181, | |
| "loss": 2.2635, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.17100633356791, | |
| "grad_norm": 0.7597271203994751, | |
| "learning_rate": 0.00011758793969849247, | |
| "loss": 2.6295, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.1738212526389866, | |
| "grad_norm": 0.770999550819397, | |
| "learning_rate": 0.00011738693467336684, | |
| "loss": 2.3357, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.1766361717100633, | |
| "grad_norm": 0.819741427898407, | |
| "learning_rate": 0.00011718592964824122, | |
| "loss": 2.4455, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.17945109078114, | |
| "grad_norm": 0.8049472570419312, | |
| "learning_rate": 0.00011698492462311558, | |
| "loss": 2.184, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.1822660098522166, | |
| "grad_norm": 0.8897677659988403, | |
| "learning_rate": 0.00011678391959798996, | |
| "loss": 2.3405, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.1850809289232935, | |
| "grad_norm": 0.9535378217697144, | |
| "learning_rate": 0.00011658291457286432, | |
| "loss": 2.3352, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.1878958479943702, | |
| "grad_norm": 0.7934727072715759, | |
| "learning_rate": 0.00011638190954773872, | |
| "loss": 2.2503, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.1907107670654469, | |
| "grad_norm": 1.1643705368041992, | |
| "learning_rate": 0.00011618090452261308, | |
| "loss": 2.4985, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.1935256861365235, | |
| "grad_norm": 1.05571448802948, | |
| "learning_rate": 0.00011597989949748745, | |
| "loss": 2.5174, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.1963406052076002, | |
| "grad_norm": 0.8346055150032043, | |
| "learning_rate": 0.00011577889447236181, | |
| "loss": 2.331, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.199155524278677, | |
| "grad_norm": 1.067415475845337, | |
| "learning_rate": 0.00011557788944723618, | |
| "loss": 2.4303, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.2019704433497538, | |
| "grad_norm": 1.0706610679626465, | |
| "learning_rate": 0.00011537688442211057, | |
| "loss": 2.2276, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.2047853624208305, | |
| "grad_norm": 0.877740204334259, | |
| "learning_rate": 0.00011517587939698493, | |
| "loss": 2.3532, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.2076002814919071, | |
| "grad_norm": 0.9245136380195618, | |
| "learning_rate": 0.0001149748743718593, | |
| "loss": 2.1982, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.2104152005629838, | |
| "grad_norm": 0.8375447392463684, | |
| "learning_rate": 0.00011477386934673366, | |
| "loss": 2.2702, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.2132301196340605, | |
| "grad_norm": 1.0361285209655762, | |
| "learning_rate": 0.00011457286432160806, | |
| "loss": 2.5355, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.2160450387051371, | |
| "grad_norm": 0.9980331063270569, | |
| "learning_rate": 0.00011437185929648242, | |
| "loss": 2.0764, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.218859957776214, | |
| "grad_norm": 0.8354774117469788, | |
| "learning_rate": 0.00011417085427135679, | |
| "loss": 2.5983, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.2216748768472907, | |
| "grad_norm": 0.8765326738357544, | |
| "learning_rate": 0.00011396984924623115, | |
| "loss": 2.0808, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.2244897959183674, | |
| "grad_norm": 1.077864408493042, | |
| "learning_rate": 0.00011376884422110554, | |
| "loss": 2.2868, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.227304714989444, | |
| "grad_norm": 0.8155612945556641, | |
| "learning_rate": 0.00011356783919597991, | |
| "loss": 2.2297, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.2301196340605207, | |
| "grad_norm": 0.9063975811004639, | |
| "learning_rate": 0.00011336683417085427, | |
| "loss": 2.522, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.2329345531315974, | |
| "grad_norm": 0.9975262880325317, | |
| "learning_rate": 0.00011316582914572864, | |
| "loss": 2.4027, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.235749472202674, | |
| "grad_norm": 0.860905110836029, | |
| "learning_rate": 0.00011296482412060303, | |
| "loss": 2.4354, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.238564391273751, | |
| "grad_norm": 0.9583187103271484, | |
| "learning_rate": 0.0001127638190954774, | |
| "loss": 2.3433, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.2413793103448276, | |
| "grad_norm": 0.8732121586799622, | |
| "learning_rate": 0.00011256281407035176, | |
| "loss": 2.3921, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.2441942294159043, | |
| "grad_norm": 0.9089124798774719, | |
| "learning_rate": 0.00011236180904522614, | |
| "loss": 2.2424, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.247009148486981, | |
| "grad_norm": 0.8566604852676392, | |
| "learning_rate": 0.00011216080402010052, | |
| "loss": 2.4264, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.2498240675580576, | |
| "grad_norm": 0.9148624539375305, | |
| "learning_rate": 0.00011195979899497488, | |
| "loss": 2.3394, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.2526389866291345, | |
| "grad_norm": 0.8913928866386414, | |
| "learning_rate": 0.00011175879396984925, | |
| "loss": 2.2236, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.255453905700211, | |
| "grad_norm": 1.1119465827941895, | |
| "learning_rate": 0.00011155778894472361, | |
| "loss": 2.391, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.2582688247712879, | |
| "grad_norm": 1.1434952020645142, | |
| "learning_rate": 0.00011135678391959799, | |
| "loss": 2.4987, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.2610837438423645, | |
| "grad_norm": 0.9885523319244385, | |
| "learning_rate": 0.00011115577889447237, | |
| "loss": 2.5184, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.2638986629134412, | |
| "grad_norm": 0.9945192337036133, | |
| "learning_rate": 0.00011095477386934675, | |
| "loss": 2.4046, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.266713581984518, | |
| "grad_norm": 0.9107452034950256, | |
| "learning_rate": 0.0001107537688442211, | |
| "loss": 2.4296, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.2695285010555946, | |
| "grad_norm": 1.2265137434005737, | |
| "learning_rate": 0.00011055276381909548, | |
| "loss": 2.4336, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.2723434201266715, | |
| "grad_norm": 0.907394289970398, | |
| "learning_rate": 0.00011035175879396986, | |
| "loss": 2.4008, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.2751583391977481, | |
| "grad_norm": 0.884708821773529, | |
| "learning_rate": 0.00011015075376884422, | |
| "loss": 2.5134, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.2779732582688248, | |
| "grad_norm": 0.8295673727989197, | |
| "learning_rate": 0.0001099497487437186, | |
| "loss": 2.5117, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.2807881773399015, | |
| "grad_norm": 1.0812764167785645, | |
| "learning_rate": 0.00010974874371859296, | |
| "loss": 2.372, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.2836030964109781, | |
| "grad_norm": 1.0535778999328613, | |
| "learning_rate": 0.00010954773869346736, | |
| "loss": 2.5114, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.2864180154820548, | |
| "grad_norm": 1.1005867719650269, | |
| "learning_rate": 0.00010934673366834172, | |
| "loss": 2.4269, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.2892329345531315, | |
| "grad_norm": 0.813443124294281, | |
| "learning_rate": 0.00010914572864321609, | |
| "loss": 2.3226, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.2920478536242084, | |
| "grad_norm": 0.8614223599433899, | |
| "learning_rate": 0.00010894472361809045, | |
| "loss": 2.3945, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.294862772695285, | |
| "grad_norm": 0.9305881857872009, | |
| "learning_rate": 0.00010874371859296483, | |
| "loss": 2.4106, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.2976776917663617, | |
| "grad_norm": 0.7936707139015198, | |
| "learning_rate": 0.00010854271356783921, | |
| "loss": 2.2523, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.3004926108374384, | |
| "grad_norm": 0.9864185452461243, | |
| "learning_rate": 0.00010834170854271357, | |
| "loss": 2.4317, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.303307529908515, | |
| "grad_norm": 0.8099750876426697, | |
| "learning_rate": 0.00010814070351758794, | |
| "loss": 2.5428, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.306122448979592, | |
| "grad_norm": 0.8694155216217041, | |
| "learning_rate": 0.00010793969849246233, | |
| "loss": 2.4084, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.3089373680506686, | |
| "grad_norm": 0.963947057723999, | |
| "learning_rate": 0.0001077386934673367, | |
| "loss": 2.3802, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.3117522871217453, | |
| "grad_norm": 0.9907119274139404, | |
| "learning_rate": 0.00010753768844221106, | |
| "loss": 2.3028, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.314567206192822, | |
| "grad_norm": 0.8978596329689026, | |
| "learning_rate": 0.00010733668341708543, | |
| "loss": 2.129, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.3173821252638986, | |
| "grad_norm": 1.0621075630187988, | |
| "learning_rate": 0.00010713567839195982, | |
| "loss": 2.4765, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.3201970443349753, | |
| "grad_norm": 1.0847358703613281, | |
| "learning_rate": 0.00010693467336683418, | |
| "loss": 2.4611, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.323011963406052, | |
| "grad_norm": 0.8706623315811157, | |
| "learning_rate": 0.00010673366834170855, | |
| "loss": 2.5458, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.3258268824771289, | |
| "grad_norm": 1.0084209442138672, | |
| "learning_rate": 0.00010653266331658291, | |
| "loss": 2.4968, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.3286418015482055, | |
| "grad_norm": 0.8770229816436768, | |
| "learning_rate": 0.00010633165829145728, | |
| "loss": 2.3268, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.3314567206192822, | |
| "grad_norm": 0.9652953743934631, | |
| "learning_rate": 0.00010613065326633167, | |
| "loss": 2.3758, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.334271639690359, | |
| "grad_norm": 0.8194919228553772, | |
| "learning_rate": 0.00010592964824120604, | |
| "loss": 2.4732, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.3370865587614356, | |
| "grad_norm": 3.534748077392578, | |
| "learning_rate": 0.0001057286432160804, | |
| "loss": 2.3725, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.3399014778325122, | |
| "grad_norm": 0.9962548017501831, | |
| "learning_rate": 0.00010552763819095478, | |
| "loss": 2.3963, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.342716396903589, | |
| "grad_norm": 0.794152021408081, | |
| "learning_rate": 0.00010532663316582916, | |
| "loss": 2.2899, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.3455313159746658, | |
| "grad_norm": 1.100648283958435, | |
| "learning_rate": 0.00010512562814070352, | |
| "loss": 2.5847, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.3483462350457425, | |
| "grad_norm": 0.8269829154014587, | |
| "learning_rate": 0.0001049246231155779, | |
| "loss": 2.4694, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.3511611541168191, | |
| "grad_norm": 0.915529727935791, | |
| "learning_rate": 0.00010472361809045225, | |
| "loss": 2.2721, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.3539760731878958, | |
| "grad_norm": 0.8491760492324829, | |
| "learning_rate": 0.00010452261306532664, | |
| "loss": 2.3008, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.3567909922589725, | |
| "grad_norm": 0.8877702355384827, | |
| "learning_rate": 0.00010432160804020101, | |
| "loss": 2.33, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.3596059113300494, | |
| "grad_norm": 0.86586993932724, | |
| "learning_rate": 0.00010412060301507539, | |
| "loss": 2.0962, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.362420830401126, | |
| "grad_norm": 0.8984941244125366, | |
| "learning_rate": 0.00010391959798994975, | |
| "loss": 2.1901, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.3652357494722027, | |
| "grad_norm": 0.8369758129119873, | |
| "learning_rate": 0.00010371859296482413, | |
| "loss": 2.4045, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.3680506685432794, | |
| "grad_norm": 0.7900081276893616, | |
| "learning_rate": 0.0001035175879396985, | |
| "loss": 2.1448, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.370865587614356, | |
| "grad_norm": 0.9296205043792725, | |
| "learning_rate": 0.00010331658291457286, | |
| "loss": 2.2958, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.3736805066854327, | |
| "grad_norm": 1.0592749118804932, | |
| "learning_rate": 0.00010311557788944724, | |
| "loss": 2.524, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.3764954257565094, | |
| "grad_norm": 0.7983985543251038, | |
| "learning_rate": 0.00010291457286432162, | |
| "loss": 2.4329, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.3793103448275863, | |
| "grad_norm": 1.1377589702606201, | |
| "learning_rate": 0.00010271356783919598, | |
| "loss": 2.2845, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.382125263898663, | |
| "grad_norm": 1.1031099557876587, | |
| "learning_rate": 0.00010251256281407036, | |
| "loss": 2.3531, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.3849401829697396, | |
| "grad_norm": 0.9376154541969299, | |
| "learning_rate": 0.00010231155778894473, | |
| "loss": 2.1406, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.3877551020408163, | |
| "grad_norm": 1.0728362798690796, | |
| "learning_rate": 0.00010211055276381909, | |
| "loss": 2.4291, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.390570021111893, | |
| "grad_norm": 1.021877408027649, | |
| "learning_rate": 0.00010190954773869348, | |
| "loss": 2.4566, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.3933849401829699, | |
| "grad_norm": 1.1455014944076538, | |
| "learning_rate": 0.00010170854271356785, | |
| "loss": 2.3074, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.3961998592540463, | |
| "grad_norm": 0.8512632846832275, | |
| "learning_rate": 0.00010150753768844221, | |
| "loss": 2.4004, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.3990147783251232, | |
| "grad_norm": 0.9220101237297058, | |
| "learning_rate": 0.00010130653266331658, | |
| "loss": 2.4456, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.4018296973962, | |
| "grad_norm": 1.0779199600219727, | |
| "learning_rate": 0.00010110552763819097, | |
| "loss": 2.4918, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.4046446164672766, | |
| "grad_norm": 1.0237290859222412, | |
| "learning_rate": 0.00010090452261306533, | |
| "loss": 2.3606, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.4074595355383532, | |
| "grad_norm": 1.2366681098937988, | |
| "learning_rate": 0.0001007035175879397, | |
| "loss": 2.468, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.41027445460943, | |
| "grad_norm": 0.9772239923477173, | |
| "learning_rate": 0.00010050251256281407, | |
| "loss": 2.2868, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.4130893736805068, | |
| "grad_norm": 0.9963237643241882, | |
| "learning_rate": 0.00010030150753768846, | |
| "loss": 2.3794, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.4159042927515835, | |
| "grad_norm": 0.8932761549949646, | |
| "learning_rate": 0.00010010050251256282, | |
| "loss": 2.4493, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.4187192118226601, | |
| "grad_norm": 0.8121969103813171, | |
| "learning_rate": 9.989949748743719e-05, | |
| "loss": 2.4045, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.4215341308937368, | |
| "grad_norm": 0.9738163948059082, | |
| "learning_rate": 9.969849246231156e-05, | |
| "loss": 2.3725, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.4243490499648135, | |
| "grad_norm": 0.8812170028686523, | |
| "learning_rate": 9.949748743718594e-05, | |
| "loss": 2.2143, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.4271639690358902, | |
| "grad_norm": 0.9500517249107361, | |
| "learning_rate": 9.929648241206031e-05, | |
| "loss": 2.4194, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.4299788881069668, | |
| "grad_norm": 0.8924652338027954, | |
| "learning_rate": 9.909547738693468e-05, | |
| "loss": 2.4061, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.4327938071780437, | |
| "grad_norm": 0.9255656003952026, | |
| "learning_rate": 9.889447236180906e-05, | |
| "loss": 2.4458, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.4356087262491204, | |
| "grad_norm": 0.9696526527404785, | |
| "learning_rate": 9.869346733668342e-05, | |
| "loss": 2.3285, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.438423645320197, | |
| "grad_norm": 0.9249640107154846, | |
| "learning_rate": 9.84924623115578e-05, | |
| "loss": 2.3348, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.4412385643912737, | |
| "grad_norm": 0.8589572906494141, | |
| "learning_rate": 9.829145728643216e-05, | |
| "loss": 2.3224, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.4440534834623504, | |
| "grad_norm": 0.9638547301292419, | |
| "learning_rate": 9.809045226130655e-05, | |
| "loss": 2.3172, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.4468684025334273, | |
| "grad_norm": 0.9466349482536316, | |
| "learning_rate": 9.788944723618091e-05, | |
| "loss": 2.4924, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.4496833216045037, | |
| "grad_norm": 0.891727089881897, | |
| "learning_rate": 9.768844221105528e-05, | |
| "loss": 2.446, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.4524982406755806, | |
| "grad_norm": 0.9032166600227356, | |
| "learning_rate": 9.748743718592965e-05, | |
| "loss": 2.3733, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.4553131597466573, | |
| "grad_norm": 0.8339729905128479, | |
| "learning_rate": 9.728643216080403e-05, | |
| "loss": 2.0429, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.458128078817734, | |
| "grad_norm": 0.8902753591537476, | |
| "learning_rate": 9.70854271356784e-05, | |
| "loss": 2.4414, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.4609429978888107, | |
| "grad_norm": 3.1321892738342285, | |
| "learning_rate": 9.688442211055276e-05, | |
| "loss": 2.1431, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.4637579169598873, | |
| "grad_norm": 0.8980495929718018, | |
| "learning_rate": 9.668341708542715e-05, | |
| "loss": 2.646, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.4665728360309642, | |
| "grad_norm": 0.9685273170471191, | |
| "learning_rate": 9.64824120603015e-05, | |
| "loss": 2.3513, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.469387755102041, | |
| "grad_norm": 0.9794145822525024, | |
| "learning_rate": 9.628140703517589e-05, | |
| "loss": 2.477, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.4722026741731176, | |
| "grad_norm": 1.0298691987991333, | |
| "learning_rate": 9.608040201005025e-05, | |
| "loss": 2.3071, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.4750175932441942, | |
| "grad_norm": 1.007864236831665, | |
| "learning_rate": 9.587939698492462e-05, | |
| "loss": 2.443, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.477832512315271, | |
| "grad_norm": 0.8953837752342224, | |
| "learning_rate": 9.5678391959799e-05, | |
| "loss": 2.645, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.4806474313863476, | |
| "grad_norm": 0.9239638447761536, | |
| "learning_rate": 9.547738693467337e-05, | |
| "loss": 2.4693, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.4834623504574243, | |
| "grad_norm": 0.9644606709480286, | |
| "learning_rate": 9.527638190954774e-05, | |
| "loss": 2.427, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.4862772695285011, | |
| "grad_norm": 0.9406245946884155, | |
| "learning_rate": 9.507537688442212e-05, | |
| "loss": 2.4382, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.4890921885995778, | |
| "grad_norm": 1.0155800580978394, | |
| "learning_rate": 9.487437185929649e-05, | |
| "loss": 2.4304, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.4919071076706545, | |
| "grad_norm": 0.8856943845748901, | |
| "learning_rate": 9.467336683417086e-05, | |
| "loss": 2.4116, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.4947220267417312, | |
| "grad_norm": 0.8708421587944031, | |
| "learning_rate": 9.447236180904523e-05, | |
| "loss": 2.4999, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.4975369458128078, | |
| "grad_norm": 0.9287678599357605, | |
| "learning_rate": 9.427135678391961e-05, | |
| "loss": 2.3666, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.5003518648838847, | |
| "grad_norm": 0.9635983109474182, | |
| "learning_rate": 9.407035175879397e-05, | |
| "loss": 2.545, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.5031667839549612, | |
| "grad_norm": 0.8631216287612915, | |
| "learning_rate": 9.386934673366835e-05, | |
| "loss": 2.3323, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.505981703026038, | |
| "grad_norm": 0.9372367858886719, | |
| "learning_rate": 9.366834170854271e-05, | |
| "loss": 2.5233, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.5087966220971147, | |
| "grad_norm": 0.8693691492080688, | |
| "learning_rate": 9.34673366834171e-05, | |
| "loss": 2.3968, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.5116115411681914, | |
| "grad_norm": 0.911888062953949, | |
| "learning_rate": 9.326633165829146e-05, | |
| "loss": 2.5019, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.514426460239268, | |
| "grad_norm": 0.9012334942817688, | |
| "learning_rate": 9.306532663316585e-05, | |
| "loss": 2.4697, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.5172413793103448, | |
| "grad_norm": 0.9713804721832275, | |
| "learning_rate": 9.28643216080402e-05, | |
| "loss": 2.4162, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.5200562983814216, | |
| "grad_norm": 0.9272815585136414, | |
| "learning_rate": 9.266331658291458e-05, | |
| "loss": 2.4306, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.522871217452498, | |
| "grad_norm": 0.9575127959251404, | |
| "learning_rate": 9.246231155778895e-05, | |
| "loss": 2.2645, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.525686136523575, | |
| "grad_norm": 1.0728868246078491, | |
| "learning_rate": 9.226130653266331e-05, | |
| "loss": 2.4701, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.5285010555946517, | |
| "grad_norm": 0.8400806784629822, | |
| "learning_rate": 9.20603015075377e-05, | |
| "loss": 2.1602, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.5313159746657283, | |
| "grad_norm": 0.9699164032936096, | |
| "learning_rate": 9.185929648241206e-05, | |
| "loss": 2.3135, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.5341308937368052, | |
| "grad_norm": 1.0054633617401123, | |
| "learning_rate": 9.165829145728644e-05, | |
| "loss": 2.2226, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.5369458128078817, | |
| "grad_norm": 0.9745274782180786, | |
| "learning_rate": 9.14572864321608e-05, | |
| "loss": 2.3635, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.5397607318789586, | |
| "grad_norm": 0.8937272429466248, | |
| "learning_rate": 9.125628140703519e-05, | |
| "loss": 2.1517, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.5425756509500352, | |
| "grad_norm": 1.0121883153915405, | |
| "learning_rate": 9.105527638190955e-05, | |
| "loss": 2.3947, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.545390570021112, | |
| "grad_norm": 0.9827173352241516, | |
| "learning_rate": 9.085427135678392e-05, | |
| "loss": 2.3363, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.5482054890921886, | |
| "grad_norm": 0.9372025728225708, | |
| "learning_rate": 9.06532663316583e-05, | |
| "loss": 2.4856, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.5510204081632653, | |
| "grad_norm": 1.1509451866149902, | |
| "learning_rate": 9.045226130653267e-05, | |
| "loss": 2.444, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.5538353272343421, | |
| "grad_norm": 1.0390241146087646, | |
| "learning_rate": 9.025125628140704e-05, | |
| "loss": 2.5131, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.5566502463054186, | |
| "grad_norm": 0.9539273381233215, | |
| "learning_rate": 9.005025125628141e-05, | |
| "loss": 2.3651, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.5594651653764955, | |
| "grad_norm": 1.0755736827850342, | |
| "learning_rate": 8.984924623115579e-05, | |
| "loss": 2.5255, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.5622800844475722, | |
| "grad_norm": 0.850516140460968, | |
| "learning_rate": 8.964824120603016e-05, | |
| "loss": 2.3374, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.5650950035186488, | |
| "grad_norm": 0.8225517868995667, | |
| "learning_rate": 8.944723618090453e-05, | |
| "loss": 2.2753, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.5679099225897255, | |
| "grad_norm": 0.9345348477363586, | |
| "learning_rate": 8.92462311557789e-05, | |
| "loss": 2.3177, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.5707248416608022, | |
| "grad_norm": 1.1513749361038208, | |
| "learning_rate": 8.904522613065326e-05, | |
| "loss": 2.5246, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.573539760731879, | |
| "grad_norm": 0.8601298332214355, | |
| "learning_rate": 8.884422110552765e-05, | |
| "loss": 2.2688, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.5763546798029555, | |
| "grad_norm": 0.9165076613426208, | |
| "learning_rate": 8.864321608040201e-05, | |
| "loss": 2.3773, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.5791695988740324, | |
| "grad_norm": 1.0467596054077148, | |
| "learning_rate": 8.84422110552764e-05, | |
| "loss": 2.3887, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.581984517945109, | |
| "grad_norm": 0.994055449962616, | |
| "learning_rate": 8.824120603015076e-05, | |
| "loss": 2.5547, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.5847994370161858, | |
| "grad_norm": 1.100698709487915, | |
| "learning_rate": 8.804020100502513e-05, | |
| "loss": 2.3517, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.5876143560872626, | |
| "grad_norm": 0.9837064743041992, | |
| "learning_rate": 8.78391959798995e-05, | |
| "loss": 2.3602, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.590429275158339, | |
| "grad_norm": 0.9642098546028137, | |
| "learning_rate": 8.763819095477387e-05, | |
| "loss": 2.3596, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.593244194229416, | |
| "grad_norm": 0.930444061756134, | |
| "learning_rate": 8.743718592964825e-05, | |
| "loss": 2.4475, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.5960591133004927, | |
| "grad_norm": 3.837113618850708, | |
| "learning_rate": 8.723618090452261e-05, | |
| "loss": 2.3049, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.5988740323715693, | |
| "grad_norm": 0.9022939205169678, | |
| "learning_rate": 8.7035175879397e-05, | |
| "loss": 2.5377, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.601688951442646, | |
| "grad_norm": 0.9081845879554749, | |
| "learning_rate": 8.683417085427135e-05, | |
| "loss": 2.2669, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.6045038705137227, | |
| "grad_norm": 1.0544410943984985, | |
| "learning_rate": 8.663316582914574e-05, | |
| "loss": 2.5433, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.6073187895847996, | |
| "grad_norm": 1.0048531293869019, | |
| "learning_rate": 8.64321608040201e-05, | |
| "loss": 2.3306, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.610133708655876, | |
| "grad_norm": 0.9438222646713257, | |
| "learning_rate": 8.623115577889449e-05, | |
| "loss": 2.3569, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.612948627726953, | |
| "grad_norm": 0.8773800134658813, | |
| "learning_rate": 8.603015075376884e-05, | |
| "loss": 2.2443, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.6157635467980296, | |
| "grad_norm": 0.8797910809516907, | |
| "learning_rate": 8.582914572864322e-05, | |
| "loss": 2.3595, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.6185784658691063, | |
| "grad_norm": 1.003212809562683, | |
| "learning_rate": 8.562814070351759e-05, | |
| "loss": 2.3623, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.621393384940183, | |
| "grad_norm": 0.97022944688797, | |
| "learning_rate": 8.542713567839196e-05, | |
| "loss": 2.3163, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.6242083040112596, | |
| "grad_norm": 1.0100817680358887, | |
| "learning_rate": 8.522613065326634e-05, | |
| "loss": 2.4039, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.6270232230823365, | |
| "grad_norm": 0.9104019403457642, | |
| "learning_rate": 8.502512562814071e-05, | |
| "loss": 2.2857, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.629838142153413, | |
| "grad_norm": 0.8757676482200623, | |
| "learning_rate": 8.482412060301508e-05, | |
| "loss": 1.8537, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.6326530612244898, | |
| "grad_norm": 1.2689822912216187, | |
| "learning_rate": 8.462311557788946e-05, | |
| "loss": 2.375, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.6354679802955665, | |
| "grad_norm": 0.9648078680038452, | |
| "learning_rate": 8.442211055276383e-05, | |
| "loss": 2.4079, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.6382828993666432, | |
| "grad_norm": 0.8941888213157654, | |
| "learning_rate": 8.42211055276382e-05, | |
| "loss": 2.3487, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.64109781843772, | |
| "grad_norm": 1.1510968208312988, | |
| "learning_rate": 8.402010050251256e-05, | |
| "loss": 2.2717, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.6439127375087965, | |
| "grad_norm": 0.8997256755828857, | |
| "learning_rate": 8.381909547738695e-05, | |
| "loss": 2.2837, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.6467276565798734, | |
| "grad_norm": 0.9518352746963501, | |
| "learning_rate": 8.36180904522613e-05, | |
| "loss": 2.4866, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.64954257565095, | |
| "grad_norm": 0.9324397444725037, | |
| "learning_rate": 8.341708542713568e-05, | |
| "loss": 2.3405, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.6523574947220268, | |
| "grad_norm": 0.9339852929115295, | |
| "learning_rate": 8.321608040201005e-05, | |
| "loss": 2.2778, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.6551724137931034, | |
| "grad_norm": 1.0167529582977295, | |
| "learning_rate": 8.301507537688443e-05, | |
| "loss": 2.3714, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.65798733286418, | |
| "grad_norm": 1.1903690099716187, | |
| "learning_rate": 8.28140703517588e-05, | |
| "loss": 2.6944, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.660802251935257, | |
| "grad_norm": 0.9818819165229797, | |
| "learning_rate": 8.261306532663317e-05, | |
| "loss": 2.5714, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.6636171710063334, | |
| "grad_norm": 0.9528331160545349, | |
| "learning_rate": 8.241206030150754e-05, | |
| "loss": 2.4357, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.6664320900774103, | |
| "grad_norm": 1.125532865524292, | |
| "learning_rate": 8.22110552763819e-05, | |
| "loss": 2.7857, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.669247009148487, | |
| "grad_norm": 1.0312976837158203, | |
| "learning_rate": 8.201005025125629e-05, | |
| "loss": 2.5711, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.6720619282195637, | |
| "grad_norm": 1.0230183601379395, | |
| "learning_rate": 8.180904522613065e-05, | |
| "loss": 2.449, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.6748768472906403, | |
| "grad_norm": 1.0071231126785278, | |
| "learning_rate": 8.160804020100504e-05, | |
| "loss": 2.4473, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.677691766361717, | |
| "grad_norm": 0.8198001980781555, | |
| "learning_rate": 8.14070351758794e-05, | |
| "loss": 2.2743, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.680506685432794, | |
| "grad_norm": 0.9489749670028687, | |
| "learning_rate": 8.120603015075378e-05, | |
| "loss": 2.3807, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.6833216045038704, | |
| "grad_norm": 0.9406233429908752, | |
| "learning_rate": 8.100502512562814e-05, | |
| "loss": 2.3478, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.6861365235749473, | |
| "grad_norm": 1.1207877397537231, | |
| "learning_rate": 8.080402010050251e-05, | |
| "loss": 2.4251, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.688951442646024, | |
| "grad_norm": 0.8841767311096191, | |
| "learning_rate": 8.060301507537689e-05, | |
| "loss": 2.4555, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.6917663617171006, | |
| "grad_norm": 1.0105196237564087, | |
| "learning_rate": 8.040201005025126e-05, | |
| "loss": 2.4965, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.6945812807881775, | |
| "grad_norm": 1.4616045951843262, | |
| "learning_rate": 8.020100502512563e-05, | |
| "loss": 2.1856, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.697396199859254, | |
| "grad_norm": 0.8796388506889343, | |
| "learning_rate": 8e-05, | |
| "loss": 2.267, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.7002111189303308, | |
| "grad_norm": 0.9459576606750488, | |
| "learning_rate": 7.979899497487438e-05, | |
| "loss": 2.3772, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.7030260380014075, | |
| "grad_norm": 0.9645008444786072, | |
| "learning_rate": 7.959798994974875e-05, | |
| "loss": 2.4959, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.7058409570724842, | |
| "grad_norm": 1.002943515777588, | |
| "learning_rate": 7.939698492462313e-05, | |
| "loss": 2.2909, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.7086558761435608, | |
| "grad_norm": 0.9787498116493225, | |
| "learning_rate": 7.91959798994975e-05, | |
| "loss": 2.4482, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.7114707952146375, | |
| "grad_norm": 0.9761974811553955, | |
| "learning_rate": 7.899497487437186e-05, | |
| "loss": 2.358, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.7142857142857144, | |
| "grad_norm": 1.0572192668914795, | |
| "learning_rate": 7.879396984924623e-05, | |
| "loss": 2.3769, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.7171006333567909, | |
| "grad_norm": 1.1789201498031616, | |
| "learning_rate": 7.85929648241206e-05, | |
| "loss": 2.4841, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.7199155524278678, | |
| "grad_norm": 1.2059478759765625, | |
| "learning_rate": 7.839195979899498e-05, | |
| "loss": 2.3738, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.7227304714989444, | |
| "grad_norm": 0.884793758392334, | |
| "learning_rate": 7.819095477386935e-05, | |
| "loss": 2.2696, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.725545390570021, | |
| "grad_norm": 1.1065027713775635, | |
| "learning_rate": 7.798994974874372e-05, | |
| "loss": 2.2802, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.7283603096410978, | |
| "grad_norm": 0.9025463461875916, | |
| "learning_rate": 7.77889447236181e-05, | |
| "loss": 2.2835, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.7311752287121744, | |
| "grad_norm": 0.939608633518219, | |
| "learning_rate": 7.758793969849247e-05, | |
| "loss": 2.2813, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.7339901477832513, | |
| "grad_norm": 1.2778400182724, | |
| "learning_rate": 7.738693467336684e-05, | |
| "loss": 2.5467, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.7368050668543278, | |
| "grad_norm": 1.1586782932281494, | |
| "learning_rate": 7.71859296482412e-05, | |
| "loss": 2.2648, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.7396199859254047, | |
| "grad_norm": 0.9858127236366272, | |
| "learning_rate": 7.698492462311559e-05, | |
| "loss": 2.053, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.7424349049964813, | |
| "grad_norm": 0.966553807258606, | |
| "learning_rate": 7.678391959798995e-05, | |
| "loss": 2.2932, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.745249824067558, | |
| "grad_norm": 1.2511391639709473, | |
| "learning_rate": 7.658291457286433e-05, | |
| "loss": 2.4427, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.748064743138635, | |
| "grad_norm": 0.9545679092407227, | |
| "learning_rate": 7.638190954773869e-05, | |
| "loss": 2.5727, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.7508796622097114, | |
| "grad_norm": 0.9427103400230408, | |
| "learning_rate": 7.618090452261307e-05, | |
| "loss": 2.213, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.7536945812807883, | |
| "grad_norm": 0.9781317710876465, | |
| "learning_rate": 7.597989949748744e-05, | |
| "loss": 2.46, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.756509500351865, | |
| "grad_norm": 0.9984252452850342, | |
| "learning_rate": 7.577889447236181e-05, | |
| "loss": 2.5281, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.7593244194229416, | |
| "grad_norm": 1.1169133186340332, | |
| "learning_rate": 7.557788944723618e-05, | |
| "loss": 2.036, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.7621393384940183, | |
| "grad_norm": 0.9950047731399536, | |
| "learning_rate": 7.537688442211056e-05, | |
| "loss": 2.3513, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.764954257565095, | |
| "grad_norm": 1.1179485321044922, | |
| "learning_rate": 7.517587939698493e-05, | |
| "loss": 2.3691, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.7677691766361718, | |
| "grad_norm": 1.3760029077529907, | |
| "learning_rate": 7.49748743718593e-05, | |
| "loss": 2.5204, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.7705840957072483, | |
| "grad_norm": 1.0102930068969727, | |
| "learning_rate": 7.477386934673368e-05, | |
| "loss": 2.4864, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.7733990147783252, | |
| "grad_norm": 0.9945108890533447, | |
| "learning_rate": 7.457286432160805e-05, | |
| "loss": 2.4329, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.7762139338494018, | |
| "grad_norm": 0.9327785968780518, | |
| "learning_rate": 7.437185929648241e-05, | |
| "loss": 2.4472, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.7790288529204785, | |
| "grad_norm": 1.280433177947998, | |
| "learning_rate": 7.417085427135678e-05, | |
| "loss": 2.5645, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.7818437719915552, | |
| "grad_norm": 0.9169235229492188, | |
| "learning_rate": 7.396984924623115e-05, | |
| "loss": 2.3657, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.7846586910626319, | |
| "grad_norm": 0.8970544338226318, | |
| "learning_rate": 7.376884422110553e-05, | |
| "loss": 2.4142, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.7874736101337088, | |
| "grad_norm": 0.9948874711990356, | |
| "learning_rate": 7.35678391959799e-05, | |
| "loss": 2.28, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.7902885292047852, | |
| "grad_norm": 1.002539873123169, | |
| "learning_rate": 7.336683417085427e-05, | |
| "loss": 2.5045, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.793103448275862, | |
| "grad_norm": 0.9109549522399902, | |
| "learning_rate": 7.316582914572865e-05, | |
| "loss": 2.5396, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.7959183673469388, | |
| "grad_norm": 1.1350212097167969, | |
| "learning_rate": 7.296482412060302e-05, | |
| "loss": 2.435, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.7987332864180154, | |
| "grad_norm": 1.1271899938583374, | |
| "learning_rate": 7.276381909547739e-05, | |
| "loss": 2.366, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.8015482054890923, | |
| "grad_norm": 0.9859978556632996, | |
| "learning_rate": 7.256281407035177e-05, | |
| "loss": 2.4066, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.8043631245601688, | |
| "grad_norm": 1.0322918891906738, | |
| "learning_rate": 7.236180904522614e-05, | |
| "loss": 2.5016, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.8071780436312457, | |
| "grad_norm": 1.045304298400879, | |
| "learning_rate": 7.21608040201005e-05, | |
| "loss": 2.332, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.8099929627023223, | |
| "grad_norm": 1.0578879117965698, | |
| "learning_rate": 7.195979899497488e-05, | |
| "loss": 2.3446, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.812807881773399, | |
| "grad_norm": 1.385239839553833, | |
| "learning_rate": 7.175879396984924e-05, | |
| "loss": 2.1476, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.8156228008444757, | |
| "grad_norm": 0.9029603600502014, | |
| "learning_rate": 7.155778894472363e-05, | |
| "loss": 2.2793, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.8184377199155524, | |
| "grad_norm": 0.9613030552864075, | |
| "learning_rate": 7.135678391959799e-05, | |
| "loss": 2.5449, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.8212526389866293, | |
| "grad_norm": 0.9739305377006531, | |
| "learning_rate": 7.115577889447236e-05, | |
| "loss": 2.3309, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.8240675580577057, | |
| "grad_norm": 1.0117180347442627, | |
| "learning_rate": 7.095477386934674e-05, | |
| "loss": 2.4027, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.8268824771287826, | |
| "grad_norm": 1.2691757678985596, | |
| "learning_rate": 7.075376884422111e-05, | |
| "loss": 2.5499, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.8296973961998593, | |
| "grad_norm": 0.9493529200553894, | |
| "learning_rate": 7.055276381909548e-05, | |
| "loss": 2.2126, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.832512315270936, | |
| "grad_norm": 1.0153197050094604, | |
| "learning_rate": 7.035175879396985e-05, | |
| "loss": 2.2784, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.8353272343420126, | |
| "grad_norm": 0.9323593974113464, | |
| "learning_rate": 7.015075376884423e-05, | |
| "loss": 2.2333, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.8381421534130893, | |
| "grad_norm": 1.0737582445144653, | |
| "learning_rate": 6.99497487437186e-05, | |
| "loss": 2.4598, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.8409570724841662, | |
| "grad_norm": 0.9445222616195679, | |
| "learning_rate": 6.974874371859297e-05, | |
| "loss": 2.2218, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.8437719915552426, | |
| "grad_norm": 1.043349027633667, | |
| "learning_rate": 6.954773869346733e-05, | |
| "loss": 2.5126, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.8465869106263195, | |
| "grad_norm": 0.9958374500274658, | |
| "learning_rate": 6.93467336683417e-05, | |
| "loss": 2.3045, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.8494018296973962, | |
| "grad_norm": 0.9490264654159546, | |
| "learning_rate": 6.914572864321608e-05, | |
| "loss": 2.3651, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.8522167487684729, | |
| "grad_norm": 1.07566499710083, | |
| "learning_rate": 6.894472361809045e-05, | |
| "loss": 2.4624, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.8550316678395498, | |
| "grad_norm": 1.0136849880218506, | |
| "learning_rate": 6.874371859296482e-05, | |
| "loss": 2.3797, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.8578465869106262, | |
| "grad_norm": 1.0830200910568237, | |
| "learning_rate": 6.85427135678392e-05, | |
| "loss": 2.4643, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.860661505981703, | |
| "grad_norm": 0.920754075050354, | |
| "learning_rate": 6.834170854271357e-05, | |
| "loss": 2.375, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.8634764250527798, | |
| "grad_norm": 1.0753567218780518, | |
| "learning_rate": 6.814070351758794e-05, | |
| "loss": 2.4276, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.8662913441238564, | |
| "grad_norm": 1.2756551504135132, | |
| "learning_rate": 6.793969849246232e-05, | |
| "loss": 2.5999, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.8691062631949331, | |
| "grad_norm": 1.165073037147522, | |
| "learning_rate": 6.773869346733669e-05, | |
| "loss": 2.5914, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.8719211822660098, | |
| "grad_norm": 1.0647106170654297, | |
| "learning_rate": 6.753768844221105e-05, | |
| "loss": 2.1331, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.8747361013370867, | |
| "grad_norm": 1.023997187614441, | |
| "learning_rate": 6.733668341708544e-05, | |
| "loss": 2.1842, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.8775510204081631, | |
| "grad_norm": 1.1140164136886597, | |
| "learning_rate": 6.71356783919598e-05, | |
| "loss": 2.3345, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.88036593947924, | |
| "grad_norm": 0.9038817882537842, | |
| "learning_rate": 6.693467336683418e-05, | |
| "loss": 2.2091, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.8831808585503167, | |
| "grad_norm": 1.0922759771347046, | |
| "learning_rate": 6.673366834170854e-05, | |
| "loss": 2.4865, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.8859957776213934, | |
| "grad_norm": 0.912775456905365, | |
| "learning_rate": 6.653266331658293e-05, | |
| "loss": 2.3812, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.88881069669247, | |
| "grad_norm": 1.0939160585403442, | |
| "learning_rate": 6.633165829145729e-05, | |
| "loss": 2.5171, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.8916256157635467, | |
| "grad_norm": 1.234376311302185, | |
| "learning_rate": 6.613065326633166e-05, | |
| "loss": 2.1198, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.8944405348346236, | |
| "grad_norm": 1.0135494470596313, | |
| "learning_rate": 6.592964824120603e-05, | |
| "loss": 2.2054, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.8972554539057, | |
| "grad_norm": 1.0782523155212402, | |
| "learning_rate": 6.57286432160804e-05, | |
| "loss": 2.4281, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.900070372976777, | |
| "grad_norm": 1.0938283205032349, | |
| "learning_rate": 6.552763819095478e-05, | |
| "loss": 2.4033, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.9028852920478536, | |
| "grad_norm": 1.080575942993164, | |
| "learning_rate": 6.532663316582915e-05, | |
| "loss": 2.2659, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.9057002111189303, | |
| "grad_norm": 0.9452334642410278, | |
| "learning_rate": 6.512562814070352e-05, | |
| "loss": 2.1833, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.9085151301900072, | |
| "grad_norm": 0.9910850524902344, | |
| "learning_rate": 6.492462311557788e-05, | |
| "loss": 2.4197, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.9113300492610836, | |
| "grad_norm": 0.9662689566612244, | |
| "learning_rate": 6.472361809045227e-05, | |
| "loss": 2.2544, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.9141449683321605, | |
| "grad_norm": 0.8501513004302979, | |
| "learning_rate": 6.452261306532663e-05, | |
| "loss": 2.3498, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.9169598874032372, | |
| "grad_norm": 1.2835460901260376, | |
| "learning_rate": 6.4321608040201e-05, | |
| "loss": 2.5167, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.9197748064743139, | |
| "grad_norm": 0.9385748505592346, | |
| "learning_rate": 6.412060301507538e-05, | |
| "loss": 2.362, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.9225897255453905, | |
| "grad_norm": 0.9999021887779236, | |
| "learning_rate": 6.391959798994975e-05, | |
| "loss": 2.2323, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.9254046446164672, | |
| "grad_norm": 1.0630273818969727, | |
| "learning_rate": 6.371859296482412e-05, | |
| "loss": 2.4489, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.928219563687544, | |
| "grad_norm": 0.9764763116836548, | |
| "learning_rate": 6.35175879396985e-05, | |
| "loss": 2.2774, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.9310344827586206, | |
| "grad_norm": 0.8416815996170044, | |
| "learning_rate": 6.331658291457287e-05, | |
| "loss": 2.2472, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.9338494018296974, | |
| "grad_norm": 0.8282995820045471, | |
| "learning_rate": 6.311557788944724e-05, | |
| "loss": 2.1936, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.9366643209007741, | |
| "grad_norm": 0.8734938502311707, | |
| "learning_rate": 6.291457286432161e-05, | |
| "loss": 2.365, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.9394792399718508, | |
| "grad_norm": 1.1187288761138916, | |
| "learning_rate": 6.271356783919599e-05, | |
| "loss": 2.3209, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.9422941590429277, | |
| "grad_norm": 0.9740754961967468, | |
| "learning_rate": 6.251256281407035e-05, | |
| "loss": 2.363, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.9451090781140041, | |
| "grad_norm": 1.023774266242981, | |
| "learning_rate": 6.231155778894473e-05, | |
| "loss": 2.2532, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.947923997185081, | |
| "grad_norm": 1.019603967666626, | |
| "learning_rate": 6.211055276381909e-05, | |
| "loss": 2.3242, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.9507389162561575, | |
| "grad_norm": 1.0288832187652588, | |
| "learning_rate": 6.190954773869348e-05, | |
| "loss": 2.6201, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.9535538353272344, | |
| "grad_norm": 0.9193139672279358, | |
| "learning_rate": 6.170854271356784e-05, | |
| "loss": 2.1794, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.956368754398311, | |
| "grad_norm": 0.9151753187179565, | |
| "learning_rate": 6.150753768844222e-05, | |
| "loss": 2.3056, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.9591836734693877, | |
| "grad_norm": 1.0923340320587158, | |
| "learning_rate": 6.130653266331658e-05, | |
| "loss": 2.2293, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.9619985925404646, | |
| "grad_norm": 0.9855085015296936, | |
| "learning_rate": 6.110552763819096e-05, | |
| "loss": 2.5036, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.964813511611541, | |
| "grad_norm": 0.9077695608139038, | |
| "learning_rate": 6.090452261306533e-05, | |
| "loss": 2.3812, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.967628430682618, | |
| "grad_norm": 0.9925841689109802, | |
| "learning_rate": 6.070351758793971e-05, | |
| "loss": 2.4093, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.9704433497536946, | |
| "grad_norm": 0.9586440324783325, | |
| "learning_rate": 6.0502512562814076e-05, | |
| "loss": 2.4851, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.9732582688247713, | |
| "grad_norm": 0.9627270698547363, | |
| "learning_rate": 6.030150753768844e-05, | |
| "loss": 2.5393, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.976073187895848, | |
| "grad_norm": 1.0779012441635132, | |
| "learning_rate": 6.0100502512562815e-05, | |
| "loss": 2.3773, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.9788881069669246, | |
| "grad_norm": 0.9611048698425293, | |
| "learning_rate": 5.989949748743718e-05, | |
| "loss": 2.3767, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.9817030260380015, | |
| "grad_norm": 0.9639135599136353, | |
| "learning_rate": 5.969849246231156e-05, | |
| "loss": 2.2936, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.984517945109078, | |
| "grad_norm": 0.9925483465194702, | |
| "learning_rate": 5.949748743718593e-05, | |
| "loss": 2.3456, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.9873328641801549, | |
| "grad_norm": 1.1674792766571045, | |
| "learning_rate": 5.929648241206031e-05, | |
| "loss": 2.3306, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.9901477832512315, | |
| "grad_norm": 1.0779776573181152, | |
| "learning_rate": 5.909547738693467e-05, | |
| "loss": 2.4675, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.9929627023223082, | |
| "grad_norm": 1.2033969163894653, | |
| "learning_rate": 5.889447236180905e-05, | |
| "loss": 2.3369, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.995777621393385, | |
| "grad_norm": 1.01941978931427, | |
| "learning_rate": 5.869346733668342e-05, | |
| "loss": 2.3271, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.9985925404644616, | |
| "grad_norm": 0.8707964420318604, | |
| "learning_rate": 5.849246231155779e-05, | |
| "loss": 2.4758, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.0014074595355384, | |
| "grad_norm": 0.9140713810920715, | |
| "learning_rate": 5.829145728643216e-05, | |
| "loss": 2.2651, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 2.004222378606615, | |
| "grad_norm": 1.016658902168274, | |
| "learning_rate": 5.809045226130654e-05, | |
| "loss": 1.7622, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 2.007037297677692, | |
| "grad_norm": 1.0154222249984741, | |
| "learning_rate": 5.7889447236180904e-05, | |
| "loss": 1.8809, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 2.0098522167487687, | |
| "grad_norm": 0.996522843837738, | |
| "learning_rate": 5.7688442211055284e-05, | |
| "loss": 2.1164, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 2.012667135819845, | |
| "grad_norm": 1.0130479335784912, | |
| "learning_rate": 5.748743718592965e-05, | |
| "loss": 2.1545, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.015482054890922, | |
| "grad_norm": 1.0692882537841797, | |
| "learning_rate": 5.728643216080403e-05, | |
| "loss": 2.0243, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 2.0182969739619985, | |
| "grad_norm": 1.0084363222122192, | |
| "learning_rate": 5.7085427135678396e-05, | |
| "loss": 1.984, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 2.0211118930330754, | |
| "grad_norm": 1.1399943828582764, | |
| "learning_rate": 5.688442211055277e-05, | |
| "loss": 1.9587, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 2.023926812104152, | |
| "grad_norm": 1.6473337411880493, | |
| "learning_rate": 5.6683417085427135e-05, | |
| "loss": 2.0001, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 2.0267417311752287, | |
| "grad_norm": 1.5070980787277222, | |
| "learning_rate": 5.6482412060301515e-05, | |
| "loss": 1.8712, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.0295566502463056, | |
| "grad_norm": 1.262854814529419, | |
| "learning_rate": 5.628140703517588e-05, | |
| "loss": 2.1427, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 2.032371569317382, | |
| "grad_norm": 1.7418184280395508, | |
| "learning_rate": 5.608040201005026e-05, | |
| "loss": 1.9851, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 2.035186488388459, | |
| "grad_norm": 1.5306885242462158, | |
| "learning_rate": 5.587939698492463e-05, | |
| "loss": 2.0147, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 2.0380014074595354, | |
| "grad_norm": 1.3408687114715576, | |
| "learning_rate": 5.567839195979899e-05, | |
| "loss": 2.1003, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 2.0408163265306123, | |
| "grad_norm": 1.3069605827331543, | |
| "learning_rate": 5.547738693467337e-05, | |
| "loss": 2.1251, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.0436312456016887, | |
| "grad_norm": 1.357084035873413, | |
| "learning_rate": 5.527638190954774e-05, | |
| "loss": 1.7681, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 2.0464461646727656, | |
| "grad_norm": 1.2578508853912354, | |
| "learning_rate": 5.507537688442211e-05, | |
| "loss": 1.8022, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 2.0492610837438425, | |
| "grad_norm": 1.2653518915176392, | |
| "learning_rate": 5.487437185929648e-05, | |
| "loss": 2.0304, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 2.052076002814919, | |
| "grad_norm": 1.2066705226898193, | |
| "learning_rate": 5.467336683417086e-05, | |
| "loss": 1.9383, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 2.054890921885996, | |
| "grad_norm": 1.2147313356399536, | |
| "learning_rate": 5.4472361809045224e-05, | |
| "loss": 1.9466, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.0577058409570723, | |
| "grad_norm": 1.2353148460388184, | |
| "learning_rate": 5.4271356783919604e-05, | |
| "loss": 2.1842, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 2.060520760028149, | |
| "grad_norm": 1.2019646167755127, | |
| "learning_rate": 5.407035175879397e-05, | |
| "loss": 1.9507, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 2.063335679099226, | |
| "grad_norm": 1.2473183870315552, | |
| "learning_rate": 5.386934673366835e-05, | |
| "loss": 1.9146, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 2.0661505981703026, | |
| "grad_norm": 1.3237521648406982, | |
| "learning_rate": 5.3668341708542716e-05, | |
| "loss": 2.026, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 2.0689655172413794, | |
| "grad_norm": 1.6444705724716187, | |
| "learning_rate": 5.346733668341709e-05, | |
| "loss": 2.1689, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.071780436312456, | |
| "grad_norm": 1.4315435886383057, | |
| "learning_rate": 5.3266331658291455e-05, | |
| "loss": 2.1981, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 2.074595355383533, | |
| "grad_norm": 1.7484960556030273, | |
| "learning_rate": 5.3065326633165835e-05, | |
| "loss": 1.998, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 2.0774102744546092, | |
| "grad_norm": 1.4129494428634644, | |
| "learning_rate": 5.28643216080402e-05, | |
| "loss": 1.9493, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 2.080225193525686, | |
| "grad_norm": 1.7426577806472778, | |
| "learning_rate": 5.266331658291458e-05, | |
| "loss": 1.8491, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 2.083040112596763, | |
| "grad_norm": 1.4977487325668335, | |
| "learning_rate": 5.246231155778895e-05, | |
| "loss": 2.0474, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.0858550316678395, | |
| "grad_norm": 1.479008674621582, | |
| "learning_rate": 5.226130653266332e-05, | |
| "loss": 2.036, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 2.0886699507389164, | |
| "grad_norm": 1.8596562147140503, | |
| "learning_rate": 5.206030150753769e-05, | |
| "loss": 2.0772, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 2.091484869809993, | |
| "grad_norm": 1.4239286184310913, | |
| "learning_rate": 5.1859296482412066e-05, | |
| "loss": 1.9441, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 2.0942997888810697, | |
| "grad_norm": 1.3117451667785645, | |
| "learning_rate": 5.165829145728643e-05, | |
| "loss": 1.8894, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 2.097114707952146, | |
| "grad_norm": 1.255926251411438, | |
| "learning_rate": 5.145728643216081e-05, | |
| "loss": 2.1199, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 2.099929627023223, | |
| "grad_norm": 1.6750807762145996, | |
| "learning_rate": 5.125628140703518e-05, | |
| "loss": 1.9147, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 2.1027445460943, | |
| "grad_norm": 1.307915210723877, | |
| "learning_rate": 5.1055276381909544e-05, | |
| "loss": 2.1592, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 2.1055594651653764, | |
| "grad_norm": 1.3630294799804688, | |
| "learning_rate": 5.0854271356783924e-05, | |
| "loss": 1.7529, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 2.1083743842364533, | |
| "grad_norm": 1.4674683809280396, | |
| "learning_rate": 5.065326633165829e-05, | |
| "loss": 1.9104, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 2.1111893033075297, | |
| "grad_norm": 1.2863240242004395, | |
| "learning_rate": 5.045226130653266e-05, | |
| "loss": 1.806, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.1140042223786066, | |
| "grad_norm": 1.4659481048583984, | |
| "learning_rate": 5.0251256281407036e-05, | |
| "loss": 2.1066, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 2.1168191414496835, | |
| "grad_norm": 1.4531869888305664, | |
| "learning_rate": 5.005025125628141e-05, | |
| "loss": 1.9145, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 2.11963406052076, | |
| "grad_norm": 1.4428577423095703, | |
| "learning_rate": 4.984924623115578e-05, | |
| "loss": 2.0346, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 2.122448979591837, | |
| "grad_norm": 1.6657663583755493, | |
| "learning_rate": 4.9648241206030155e-05, | |
| "loss": 1.8736, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 2.1252638986629133, | |
| "grad_norm": 1.45827317237854, | |
| "learning_rate": 4.944723618090453e-05, | |
| "loss": 1.8692, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.12807881773399, | |
| "grad_norm": 1.4714118242263794, | |
| "learning_rate": 4.92462311557789e-05, | |
| "loss": 2.1693, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 2.1308937368050667, | |
| "grad_norm": 1.4159564971923828, | |
| "learning_rate": 4.9045226130653274e-05, | |
| "loss": 2.0312, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 2.1337086558761436, | |
| "grad_norm": 1.409199833869934, | |
| "learning_rate": 4.884422110552764e-05, | |
| "loss": 2.043, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 2.1365235749472204, | |
| "grad_norm": 1.4486503601074219, | |
| "learning_rate": 4.864321608040201e-05, | |
| "loss": 1.8924, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 2.139338494018297, | |
| "grad_norm": 1.5751312971115112, | |
| "learning_rate": 4.844221105527638e-05, | |
| "loss": 1.8043, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.142153413089374, | |
| "grad_norm": 1.4712185859680176, | |
| "learning_rate": 4.824120603015075e-05, | |
| "loss": 1.9763, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 2.1449683321604502, | |
| "grad_norm": 1.3530285358428955, | |
| "learning_rate": 4.8040201005025125e-05, | |
| "loss": 1.7487, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 2.147783251231527, | |
| "grad_norm": 1.5095936059951782, | |
| "learning_rate": 4.78391959798995e-05, | |
| "loss": 2.0177, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 2.1505981703026036, | |
| "grad_norm": 1.2347254753112793, | |
| "learning_rate": 4.763819095477387e-05, | |
| "loss": 1.9277, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 2.1534130893736805, | |
| "grad_norm": 1.728926658630371, | |
| "learning_rate": 4.7437185929648244e-05, | |
| "loss": 2.0394, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.1562280084447574, | |
| "grad_norm": 1.3105862140655518, | |
| "learning_rate": 4.723618090452262e-05, | |
| "loss": 1.9132, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 2.159042927515834, | |
| "grad_norm": 1.4253538846969604, | |
| "learning_rate": 4.703517587939698e-05, | |
| "loss": 1.7072, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 2.1618578465869107, | |
| "grad_norm": 1.5160298347473145, | |
| "learning_rate": 4.6834170854271356e-05, | |
| "loss": 2.3564, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 2.164672765657987, | |
| "grad_norm": 1.384318470954895, | |
| "learning_rate": 4.663316582914573e-05, | |
| "loss": 1.911, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 2.167487684729064, | |
| "grad_norm": 1.6801820993423462, | |
| "learning_rate": 4.64321608040201e-05, | |
| "loss": 2.0506, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.170302603800141, | |
| "grad_norm": 1.401426076889038, | |
| "learning_rate": 4.6231155778894475e-05, | |
| "loss": 1.6387, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 2.1731175228712174, | |
| "grad_norm": 1.4150290489196777, | |
| "learning_rate": 4.603015075376885e-05, | |
| "loss": 2.1757, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 2.1759324419422943, | |
| "grad_norm": 1.909029483795166, | |
| "learning_rate": 4.582914572864322e-05, | |
| "loss": 1.7829, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 2.1787473610133707, | |
| "grad_norm": 1.530287265777588, | |
| "learning_rate": 4.5628140703517594e-05, | |
| "loss": 2.0015, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 2.1815622800844476, | |
| "grad_norm": 1.5270819664001465, | |
| "learning_rate": 4.542713567839196e-05, | |
| "loss": 2.1369, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.184377199155524, | |
| "grad_norm": 2.737819194793701, | |
| "learning_rate": 4.522613065326633e-05, | |
| "loss": 1.7966, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 2.187192118226601, | |
| "grad_norm": 1.4319558143615723, | |
| "learning_rate": 4.5025125628140706e-05, | |
| "loss": 1.949, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 2.190007037297678, | |
| "grad_norm": 1.5846929550170898, | |
| "learning_rate": 4.482412060301508e-05, | |
| "loss": 1.9676, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 2.1928219563687543, | |
| "grad_norm": 1.3088924884796143, | |
| "learning_rate": 4.462311557788945e-05, | |
| "loss": 1.9452, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 2.195636875439831, | |
| "grad_norm": 1.4991919994354248, | |
| "learning_rate": 4.4422110552763825e-05, | |
| "loss": 1.8966, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.1984517945109077, | |
| "grad_norm": 1.306575059890747, | |
| "learning_rate": 4.42211055276382e-05, | |
| "loss": 1.9106, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 2.2012667135819846, | |
| "grad_norm": 1.562092900276184, | |
| "learning_rate": 4.4020100502512564e-05, | |
| "loss": 1.8167, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 2.204081632653061, | |
| "grad_norm": 1.6543974876403809, | |
| "learning_rate": 4.381909547738694e-05, | |
| "loss": 1.9092, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 2.206896551724138, | |
| "grad_norm": 1.485269546508789, | |
| "learning_rate": 4.3618090452261303e-05, | |
| "loss": 1.9781, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 2.209711470795215, | |
| "grad_norm": 1.2385632991790771, | |
| "learning_rate": 4.3417085427135676e-05, | |
| "loss": 1.87, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.2125263898662912, | |
| "grad_norm": 1.2291756868362427, | |
| "learning_rate": 4.321608040201005e-05, | |
| "loss": 2.0753, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 2.215341308937368, | |
| "grad_norm": 1.3407044410705566, | |
| "learning_rate": 4.301507537688442e-05, | |
| "loss": 1.8011, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 2.2181562280084446, | |
| "grad_norm": 1.5226972103118896, | |
| "learning_rate": 4.2814070351758795e-05, | |
| "loss": 2.031, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 2.2209711470795215, | |
| "grad_norm": 1.2508612871170044, | |
| "learning_rate": 4.261306532663317e-05, | |
| "loss": 1.9893, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 2.2237860661505984, | |
| "grad_norm": 1.8101375102996826, | |
| "learning_rate": 4.241206030150754e-05, | |
| "loss": 2.0614, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.226600985221675, | |
| "grad_norm": 1.3492687940597534, | |
| "learning_rate": 4.2211055276381914e-05, | |
| "loss": 1.5472, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 2.2294159042927517, | |
| "grad_norm": 1.4373085498809814, | |
| "learning_rate": 4.201005025125628e-05, | |
| "loss": 1.9417, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 2.232230823363828, | |
| "grad_norm": 1.343981385231018, | |
| "learning_rate": 4.180904522613065e-05, | |
| "loss": 1.9317, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 2.235045742434905, | |
| "grad_norm": 1.3740363121032715, | |
| "learning_rate": 4.1608040201005026e-05, | |
| "loss": 1.7677, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 2.2378606615059815, | |
| "grad_norm": 1.4676454067230225, | |
| "learning_rate": 4.14070351758794e-05, | |
| "loss": 1.9661, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.2406755805770584, | |
| "grad_norm": 1.320854902267456, | |
| "learning_rate": 4.120603015075377e-05, | |
| "loss": 1.8218, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 2.2434904996481353, | |
| "grad_norm": 1.7027606964111328, | |
| "learning_rate": 4.1005025125628145e-05, | |
| "loss": 1.979, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 2.2463054187192117, | |
| "grad_norm": 1.363239049911499, | |
| "learning_rate": 4.080402010050252e-05, | |
| "loss": 1.8902, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 2.2491203377902886, | |
| "grad_norm": 1.707664966583252, | |
| "learning_rate": 4.060301507537689e-05, | |
| "loss": 2.0026, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 2.251935256861365, | |
| "grad_norm": 1.7282025814056396, | |
| "learning_rate": 4.040201005025126e-05, | |
| "loss": 1.8098, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.254750175932442, | |
| "grad_norm": 1.5891460180282593, | |
| "learning_rate": 4.020100502512563e-05, | |
| "loss": 1.8639, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 2.2575650950035184, | |
| "grad_norm": 1.5569334030151367, | |
| "learning_rate": 4e-05, | |
| "loss": 2.1027, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 2.2603800140745953, | |
| "grad_norm": 1.4195587635040283, | |
| "learning_rate": 3.9798994974874376e-05, | |
| "loss": 1.8757, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 2.263194933145672, | |
| "grad_norm": 1.3400124311447144, | |
| "learning_rate": 3.959798994974875e-05, | |
| "loss": 1.943, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 2.2660098522167487, | |
| "grad_norm": 1.4379513263702393, | |
| "learning_rate": 3.9396984924623115e-05, | |
| "loss": 2.0351, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 2.2688247712878256, | |
| "grad_norm": 1.8557440042495728, | |
| "learning_rate": 3.919597989949749e-05, | |
| "loss": 1.6522, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 2.271639690358902, | |
| "grad_norm": 1.68703031539917, | |
| "learning_rate": 3.899497487437186e-05, | |
| "loss": 1.8939, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 2.274454609429979, | |
| "grad_norm": 1.4797513484954834, | |
| "learning_rate": 3.8793969849246234e-05, | |
| "loss": 1.7212, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 2.277269528501056, | |
| "grad_norm": 2.070215940475464, | |
| "learning_rate": 3.85929648241206e-05, | |
| "loss": 1.9758, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 2.2800844475721322, | |
| "grad_norm": 1.4958938360214233, | |
| "learning_rate": 3.8391959798994973e-05, | |
| "loss": 2.0214, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.282899366643209, | |
| "grad_norm": 1.4052972793579102, | |
| "learning_rate": 3.8190954773869346e-05, | |
| "loss": 1.9959, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 2.2857142857142856, | |
| "grad_norm": 1.652631402015686, | |
| "learning_rate": 3.798994974874372e-05, | |
| "loss": 1.8977, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 2.2885292047853625, | |
| "grad_norm": 1.4963494539260864, | |
| "learning_rate": 3.778894472361809e-05, | |
| "loss": 1.7929, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 2.2913441238564394, | |
| "grad_norm": 1.554140329360962, | |
| "learning_rate": 3.7587939698492465e-05, | |
| "loss": 1.9422, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 2.294159042927516, | |
| "grad_norm": 1.5336120128631592, | |
| "learning_rate": 3.738693467336684e-05, | |
| "loss": 2.0812, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.2969739619985927, | |
| "grad_norm": 1.998458981513977, | |
| "learning_rate": 3.7185929648241204e-05, | |
| "loss": 1.8639, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 2.299788881069669, | |
| "grad_norm": 1.662591814994812, | |
| "learning_rate": 3.698492462311558e-05, | |
| "loss": 2.0354, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 2.302603800140746, | |
| "grad_norm": 1.6507760286331177, | |
| "learning_rate": 3.678391959798995e-05, | |
| "loss": 2.0235, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 2.3054187192118225, | |
| "grad_norm": 1.6057195663452148, | |
| "learning_rate": 3.658291457286432e-05, | |
| "loss": 1.9731, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 2.3082336382828994, | |
| "grad_norm": 1.5184822082519531, | |
| "learning_rate": 3.6381909547738696e-05, | |
| "loss": 1.7881, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.311048557353976, | |
| "grad_norm": 1.5405902862548828, | |
| "learning_rate": 3.618090452261307e-05, | |
| "loss": 1.9947, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 2.3138634764250527, | |
| "grad_norm": 1.880598545074463, | |
| "learning_rate": 3.597989949748744e-05, | |
| "loss": 1.9817, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 2.3166783954961296, | |
| "grad_norm": 1.707992434501648, | |
| "learning_rate": 3.5778894472361815e-05, | |
| "loss": 1.7769, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 2.319493314567206, | |
| "grad_norm": 1.4633187055587769, | |
| "learning_rate": 3.557788944723618e-05, | |
| "loss": 1.8575, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 2.322308233638283, | |
| "grad_norm": 1.4804046154022217, | |
| "learning_rate": 3.5376884422110554e-05, | |
| "loss": 1.9168, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.3251231527093594, | |
| "grad_norm": 1.6535083055496216, | |
| "learning_rate": 3.517587939698493e-05, | |
| "loss": 2.0092, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 2.3279380717804363, | |
| "grad_norm": 1.45828115940094, | |
| "learning_rate": 3.49748743718593e-05, | |
| "loss": 1.824, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 2.330752990851513, | |
| "grad_norm": 1.4650769233703613, | |
| "learning_rate": 3.4773869346733667e-05, | |
| "loss": 2.0699, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 2.3335679099225897, | |
| "grad_norm": 1.629009485244751, | |
| "learning_rate": 3.457286432160804e-05, | |
| "loss": 1.856, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 2.3363828289936666, | |
| "grad_norm": 1.6346815824508667, | |
| "learning_rate": 3.437185929648241e-05, | |
| "loss": 1.7844, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.339197748064743, | |
| "grad_norm": 2.298454523086548, | |
| "learning_rate": 3.4170854271356785e-05, | |
| "loss": 1.8711, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 2.34201266713582, | |
| "grad_norm": 1.6962236166000366, | |
| "learning_rate": 3.396984924623116e-05, | |
| "loss": 1.9879, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 2.344827586206897, | |
| "grad_norm": 1.5078356266021729, | |
| "learning_rate": 3.3768844221105525e-05, | |
| "loss": 1.6655, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 2.3476425052779732, | |
| "grad_norm": 1.9207500219345093, | |
| "learning_rate": 3.35678391959799e-05, | |
| "loss": 1.7975, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 2.35045742434905, | |
| "grad_norm": 1.5677213668823242, | |
| "learning_rate": 3.336683417085427e-05, | |
| "loss": 2.0483, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 2.3532723434201266, | |
| "grad_norm": 1.6488611698150635, | |
| "learning_rate": 3.3165829145728643e-05, | |
| "loss": 1.8616, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 2.3560872624912035, | |
| "grad_norm": 1.718641996383667, | |
| "learning_rate": 3.2964824120603016e-05, | |
| "loss": 1.9279, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 2.35890218156228, | |
| "grad_norm": 1.6529680490493774, | |
| "learning_rate": 3.276381909547739e-05, | |
| "loss": 2.1128, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 2.361717100633357, | |
| "grad_norm": 1.691375732421875, | |
| "learning_rate": 3.256281407035176e-05, | |
| "loss": 1.801, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 2.3645320197044333, | |
| "grad_norm": 1.6755695343017578, | |
| "learning_rate": 3.2361809045226135e-05, | |
| "loss": 2.0209, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.36734693877551, | |
| "grad_norm": 1.3565911054611206, | |
| "learning_rate": 3.21608040201005e-05, | |
| "loss": 1.8019, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 2.370161857846587, | |
| "grad_norm": 1.4155783653259277, | |
| "learning_rate": 3.1959798994974875e-05, | |
| "loss": 1.9121, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 2.3729767769176635, | |
| "grad_norm": 1.6148691177368164, | |
| "learning_rate": 3.175879396984925e-05, | |
| "loss": 1.9028, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 2.3757916959887404, | |
| "grad_norm": 1.408504605293274, | |
| "learning_rate": 3.155778894472362e-05, | |
| "loss": 1.7751, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 2.378606615059817, | |
| "grad_norm": 1.4195948839187622, | |
| "learning_rate": 3.1356783919597993e-05, | |
| "loss": 1.8083, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 2.3814215341308937, | |
| "grad_norm": 1.5938619375228882, | |
| "learning_rate": 3.1155778894472366e-05, | |
| "loss": 1.8096, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 2.3842364532019706, | |
| "grad_norm": 1.4775474071502686, | |
| "learning_rate": 3.095477386934674e-05, | |
| "loss": 2.0482, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 2.387051372273047, | |
| "grad_norm": 1.3416311740875244, | |
| "learning_rate": 3.075376884422111e-05, | |
| "loss": 1.8107, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 2.389866291344124, | |
| "grad_norm": 1.6304892301559448, | |
| "learning_rate": 3.055276381909548e-05, | |
| "loss": 1.913, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 2.3926812104152004, | |
| "grad_norm": 1.2348568439483643, | |
| "learning_rate": 3.0351758793969855e-05, | |
| "loss": 1.8254, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.3954961294862773, | |
| "grad_norm": 1.6292930841445923, | |
| "learning_rate": 3.015075376884422e-05, | |
| "loss": 2.1953, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 2.398311048557354, | |
| "grad_norm": 1.6192723512649536, | |
| "learning_rate": 2.994974874371859e-05, | |
| "loss": 1.7694, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 2.4011259676284307, | |
| "grad_norm": 1.6231796741485596, | |
| "learning_rate": 2.9748743718592964e-05, | |
| "loss": 2.1208, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 2.4039408866995076, | |
| "grad_norm": 1.3113828897476196, | |
| "learning_rate": 2.9547738693467337e-05, | |
| "loss": 1.9544, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 2.406755805770584, | |
| "grad_norm": 1.6705840826034546, | |
| "learning_rate": 2.934673366834171e-05, | |
| "loss": 2.0316, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 2.409570724841661, | |
| "grad_norm": 1.7489991188049316, | |
| "learning_rate": 2.914572864321608e-05, | |
| "loss": 2.1702, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 2.4123856439127374, | |
| "grad_norm": 1.7634392976760864, | |
| "learning_rate": 2.8944723618090452e-05, | |
| "loss": 2.003, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 2.4152005629838142, | |
| "grad_norm": 1.607228398323059, | |
| "learning_rate": 2.8743718592964825e-05, | |
| "loss": 1.9383, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 2.4180154820548907, | |
| "grad_norm": 1.661271095275879, | |
| "learning_rate": 2.8542713567839198e-05, | |
| "loss": 1.9874, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 2.4208304011259676, | |
| "grad_norm": 1.4608184099197388, | |
| "learning_rate": 2.8341708542713568e-05, | |
| "loss": 1.8086, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.4236453201970445, | |
| "grad_norm": 1.4614999294281006, | |
| "learning_rate": 2.814070351758794e-05, | |
| "loss": 2.069, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 2.426460239268121, | |
| "grad_norm": 1.5575437545776367, | |
| "learning_rate": 2.7939698492462314e-05, | |
| "loss": 1.8134, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 2.429275158339198, | |
| "grad_norm": 1.5272866487503052, | |
| "learning_rate": 2.7738693467336686e-05, | |
| "loss": 1.9724, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 2.4320900774102743, | |
| "grad_norm": 1.524636149406433, | |
| "learning_rate": 2.7537688442211056e-05, | |
| "loss": 1.8841, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 2.434904996481351, | |
| "grad_norm": 1.7057536840438843, | |
| "learning_rate": 2.733668341708543e-05, | |
| "loss": 1.9102, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 2.437719915552428, | |
| "grad_norm": 1.4803720712661743, | |
| "learning_rate": 2.7135678391959802e-05, | |
| "loss": 1.8558, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 2.4405348346235045, | |
| "grad_norm": 1.486907958984375, | |
| "learning_rate": 2.6934673366834175e-05, | |
| "loss": 2.0651, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 2.4433497536945814, | |
| "grad_norm": 1.5045924186706543, | |
| "learning_rate": 2.6733668341708545e-05, | |
| "loss": 1.8523, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 2.446164672765658, | |
| "grad_norm": 1.5075145959854126, | |
| "learning_rate": 2.6532663316582917e-05, | |
| "loss": 2.1232, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 2.4489795918367347, | |
| "grad_norm": 1.5650744438171387, | |
| "learning_rate": 2.633165829145729e-05, | |
| "loss": 1.9524, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.4517945109078116, | |
| "grad_norm": 1.4480630159378052, | |
| "learning_rate": 2.613065326633166e-05, | |
| "loss": 1.8816, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 2.454609429978888, | |
| "grad_norm": 1.7436559200286865, | |
| "learning_rate": 2.5929648241206033e-05, | |
| "loss": 2.0141, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 2.457424349049965, | |
| "grad_norm": 1.647824764251709, | |
| "learning_rate": 2.5728643216080406e-05, | |
| "loss": 1.9715, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 2.4602392681210414, | |
| "grad_norm": 1.5032564401626587, | |
| "learning_rate": 2.5527638190954772e-05, | |
| "loss": 1.9423, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 2.4630541871921183, | |
| "grad_norm": 1.615399718284607, | |
| "learning_rate": 2.5326633165829145e-05, | |
| "loss": 2.1547, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.4658691062631948, | |
| "grad_norm": 1.66806161403656, | |
| "learning_rate": 2.5125628140703518e-05, | |
| "loss": 2.0234, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 2.4686840253342717, | |
| "grad_norm": 1.69028902053833, | |
| "learning_rate": 2.492462311557789e-05, | |
| "loss": 1.9796, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 2.471498944405348, | |
| "grad_norm": 1.5616704225540161, | |
| "learning_rate": 2.4723618090452264e-05, | |
| "loss": 1.979, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 2.474313863476425, | |
| "grad_norm": 1.9459314346313477, | |
| "learning_rate": 2.4522613065326637e-05, | |
| "loss": 2.0534, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 2.477128782547502, | |
| "grad_norm": 1.2826955318450928, | |
| "learning_rate": 2.4321608040201007e-05, | |
| "loss": 1.9959, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.4799437016185784, | |
| "grad_norm": 1.3462079763412476, | |
| "learning_rate": 2.4120603015075376e-05, | |
| "loss": 1.7451, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 2.4827586206896552, | |
| "grad_norm": 1.7991423606872559, | |
| "learning_rate": 2.391959798994975e-05, | |
| "loss": 2.1944, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 2.4855735397607317, | |
| "grad_norm": 1.4024704694747925, | |
| "learning_rate": 2.3718592964824122e-05, | |
| "loss": 1.9266, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 2.4883884588318086, | |
| "grad_norm": 1.7428147792816162, | |
| "learning_rate": 2.351758793969849e-05, | |
| "loss": 1.991, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 2.4912033779028855, | |
| "grad_norm": 1.4942609071731567, | |
| "learning_rate": 2.3316582914572865e-05, | |
| "loss": 1.7034, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 2.494018296973962, | |
| "grad_norm": 1.6050865650177002, | |
| "learning_rate": 2.3115577889447238e-05, | |
| "loss": 1.9718, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 2.496833216045039, | |
| "grad_norm": 1.6679102182388306, | |
| "learning_rate": 2.291457286432161e-05, | |
| "loss": 1.7724, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 2.4996481351161153, | |
| "grad_norm": 1.4811137914657593, | |
| "learning_rate": 2.271356783919598e-05, | |
| "loss": 1.9245, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 2.502463054187192, | |
| "grad_norm": 1.5758980512619019, | |
| "learning_rate": 2.2512562814070353e-05, | |
| "loss": 1.9094, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 2.505277973258269, | |
| "grad_norm": 1.6346875429153442, | |
| "learning_rate": 2.2311557788944726e-05, | |
| "loss": 1.8482, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.5080928923293455, | |
| "grad_norm": 1.6329705715179443, | |
| "learning_rate": 2.21105527638191e-05, | |
| "loss": 1.9275, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 2.510907811400422, | |
| "grad_norm": 1.9860655069351196, | |
| "learning_rate": 2.190954773869347e-05, | |
| "loss": 2.0708, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 2.513722730471499, | |
| "grad_norm": 2.201899528503418, | |
| "learning_rate": 2.1708542713567838e-05, | |
| "loss": 1.5935, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 2.5165376495425757, | |
| "grad_norm": 1.7361814975738525, | |
| "learning_rate": 2.150753768844221e-05, | |
| "loss": 1.716, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 2.519352568613652, | |
| "grad_norm": 1.6891804933547974, | |
| "learning_rate": 2.1306532663316584e-05, | |
| "loss": 2.0822, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 2.522167487684729, | |
| "grad_norm": 1.5002251863479614, | |
| "learning_rate": 2.1105527638190957e-05, | |
| "loss": 1.8622, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 2.5249824067558055, | |
| "grad_norm": 1.6818735599517822, | |
| "learning_rate": 2.0904522613065327e-05, | |
| "loss": 1.7687, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 2.5277973258268824, | |
| "grad_norm": 1.6019138097763062, | |
| "learning_rate": 2.07035175879397e-05, | |
| "loss": 1.8481, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 2.5306122448979593, | |
| "grad_norm": 1.517175555229187, | |
| "learning_rate": 2.0502512562814073e-05, | |
| "loss": 1.8791, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 2.533427163969036, | |
| "grad_norm": 1.4796918630599976, | |
| "learning_rate": 2.0301507537688446e-05, | |
| "loss": 1.8031, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.5362420830401127, | |
| "grad_norm": 1.5934321880340576, | |
| "learning_rate": 2.0100502512562815e-05, | |
| "loss": 1.9215, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 2.539057002111189, | |
| "grad_norm": 1.5581581592559814, | |
| "learning_rate": 1.9899497487437188e-05, | |
| "loss": 1.9739, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 2.541871921182266, | |
| "grad_norm": 1.6254914999008179, | |
| "learning_rate": 1.9698492462311558e-05, | |
| "loss": 2.0283, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 2.544686840253343, | |
| "grad_norm": 1.5393351316452026, | |
| "learning_rate": 1.949748743718593e-05, | |
| "loss": 2.1366, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.5475017593244194, | |
| "grad_norm": 1.369229793548584, | |
| "learning_rate": 1.92964824120603e-05, | |
| "loss": 1.8624, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.5503166783954963, | |
| "grad_norm": 1.5214154720306396, | |
| "learning_rate": 1.9095477386934673e-05, | |
| "loss": 1.8812, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 2.5531315974665727, | |
| "grad_norm": 1.7752878665924072, | |
| "learning_rate": 1.8894472361809046e-05, | |
| "loss": 1.959, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 2.5559465165376496, | |
| "grad_norm": 1.5053621530532837, | |
| "learning_rate": 1.869346733668342e-05, | |
| "loss": 1.895, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.5587614356087265, | |
| "grad_norm": 1.660584568977356, | |
| "learning_rate": 1.849246231155779e-05, | |
| "loss": 1.8461, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 2.561576354679803, | |
| "grad_norm": 1.5243057012557983, | |
| "learning_rate": 1.829145728643216e-05, | |
| "loss": 1.7571, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.5643912737508794, | |
| "grad_norm": 1.4486278295516968, | |
| "learning_rate": 1.8090452261306535e-05, | |
| "loss": 1.9905, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 2.5672061928219563, | |
| "grad_norm": 1.5734102725982666, | |
| "learning_rate": 1.7889447236180908e-05, | |
| "loss": 1.8319, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.570021111893033, | |
| "grad_norm": 1.524849534034729, | |
| "learning_rate": 1.7688442211055277e-05, | |
| "loss": 1.8446, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 2.5728360309641096, | |
| "grad_norm": 1.4271085262298584, | |
| "learning_rate": 1.748743718592965e-05, | |
| "loss": 1.932, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.5756509500351865, | |
| "grad_norm": 1.4514641761779785, | |
| "learning_rate": 1.728643216080402e-05, | |
| "loss": 1.8912, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.578465869106263, | |
| "grad_norm": 1.5679149627685547, | |
| "learning_rate": 1.7085427135678393e-05, | |
| "loss": 1.8389, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.58128078817734, | |
| "grad_norm": 1.628262996673584, | |
| "learning_rate": 1.6884422110552762e-05, | |
| "loss": 1.7108, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 2.5840957072484168, | |
| "grad_norm": 1.466387152671814, | |
| "learning_rate": 1.6683417085427135e-05, | |
| "loss": 1.7445, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.586910626319493, | |
| "grad_norm": 1.6148653030395508, | |
| "learning_rate": 1.6482412060301508e-05, | |
| "loss": 1.8271, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 2.58972554539057, | |
| "grad_norm": 1.6727656126022339, | |
| "learning_rate": 1.628140703517588e-05, | |
| "loss": 1.8221, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.5925404644616465, | |
| "grad_norm": 1.6274527311325073, | |
| "learning_rate": 1.608040201005025e-05, | |
| "loss": 1.9275, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 2.5953553835327234, | |
| "grad_norm": 1.5122441053390503, | |
| "learning_rate": 1.5879396984924624e-05, | |
| "loss": 1.7531, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.5981703026038003, | |
| "grad_norm": 1.5030601024627686, | |
| "learning_rate": 1.5678391959798997e-05, | |
| "loss": 1.9965, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 2.600985221674877, | |
| "grad_norm": 1.7044039964675903, | |
| "learning_rate": 1.547738693467337e-05, | |
| "loss": 1.946, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.6038001407459537, | |
| "grad_norm": 1.505894422531128, | |
| "learning_rate": 1.527638190954774e-05, | |
| "loss": 1.8394, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.60661505981703, | |
| "grad_norm": 1.5264232158660889, | |
| "learning_rate": 1.507537688442211e-05, | |
| "loss": 1.8762, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 2.609429978888107, | |
| "grad_norm": 1.512060284614563, | |
| "learning_rate": 1.4874371859296482e-05, | |
| "loss": 1.9039, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 2.612244897959184, | |
| "grad_norm": 1.6046111583709717, | |
| "learning_rate": 1.4673366834170855e-05, | |
| "loss": 1.8746, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 2.6150598170302604, | |
| "grad_norm": 1.7210888862609863, | |
| "learning_rate": 1.4472361809045226e-05, | |
| "loss": 1.8613, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 2.6178747361013373, | |
| "grad_norm": 1.7266684770584106, | |
| "learning_rate": 1.4271356783919599e-05, | |
| "loss": 1.966, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.6206896551724137, | |
| "grad_norm": 1.6090869903564453, | |
| "learning_rate": 1.407035175879397e-05, | |
| "loss": 1.8954, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 2.6235045742434906, | |
| "grad_norm": 1.6034605503082275, | |
| "learning_rate": 1.3869346733668343e-05, | |
| "loss": 1.9799, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 2.626319493314567, | |
| "grad_norm": 1.8251813650131226, | |
| "learning_rate": 1.3668341708542715e-05, | |
| "loss": 1.8237, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 2.629134412385644, | |
| "grad_norm": 1.9154014587402344, | |
| "learning_rate": 1.3467336683417087e-05, | |
| "loss": 1.7706, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 2.6319493314567204, | |
| "grad_norm": 1.6452045440673828, | |
| "learning_rate": 1.3266331658291459e-05, | |
| "loss": 2.212, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.6347642505277973, | |
| "grad_norm": 1.5571489334106445, | |
| "learning_rate": 1.306532663316583e-05, | |
| "loss": 1.7057, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 2.637579169598874, | |
| "grad_norm": 1.5120084285736084, | |
| "learning_rate": 1.2864321608040203e-05, | |
| "loss": 1.958, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 2.6403940886699506, | |
| "grad_norm": 1.4263646602630615, | |
| "learning_rate": 1.2663316582914573e-05, | |
| "loss": 2.2181, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 2.6432090077410275, | |
| "grad_norm": 1.7924742698669434, | |
| "learning_rate": 1.2462311557788946e-05, | |
| "loss": 2.0682, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 2.646023926812104, | |
| "grad_norm": 1.568236231803894, | |
| "learning_rate": 1.2261306532663318e-05, | |
| "loss": 1.9372, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.648838845883181, | |
| "grad_norm": 1.826130986213684, | |
| "learning_rate": 1.2060301507537688e-05, | |
| "loss": 1.8557, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 2.6516537649542578, | |
| "grad_norm": 1.4434233903884888, | |
| "learning_rate": 1.1859296482412061e-05, | |
| "loss": 1.7765, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 2.654468684025334, | |
| "grad_norm": 1.4508579969406128, | |
| "learning_rate": 1.1658291457286432e-05, | |
| "loss": 1.9047, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 2.657283603096411, | |
| "grad_norm": 1.6055655479431152, | |
| "learning_rate": 1.1457286432160805e-05, | |
| "loss": 1.9673, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 2.6600985221674875, | |
| "grad_norm": 1.7329376935958862, | |
| "learning_rate": 1.1256281407035177e-05, | |
| "loss": 1.9288, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.6629134412385644, | |
| "grad_norm": 2.1003239154815674, | |
| "learning_rate": 1.105527638190955e-05, | |
| "loss": 1.9618, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 2.6657283603096413, | |
| "grad_norm": 1.5641478300094604, | |
| "learning_rate": 1.0854271356783919e-05, | |
| "loss": 1.9631, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 2.668543279380718, | |
| "grad_norm": 1.5357648134231567, | |
| "learning_rate": 1.0653266331658292e-05, | |
| "loss": 2.0265, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 2.6713581984517947, | |
| "grad_norm": 1.3177186250686646, | |
| "learning_rate": 1.0452261306532663e-05, | |
| "loss": 1.878, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 2.674173117522871, | |
| "grad_norm": 1.499448299407959, | |
| "learning_rate": 1.0251256281407036e-05, | |
| "loss": 1.8745, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.676988036593948, | |
| "grad_norm": 1.6168919801712036, | |
| "learning_rate": 1.0050251256281408e-05, | |
| "loss": 2.0069, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 2.6798029556650245, | |
| "grad_norm": 1.6026453971862793, | |
| "learning_rate": 9.849246231155779e-06, | |
| "loss": 1.7609, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 2.6826178747361014, | |
| "grad_norm": 1.796265721321106, | |
| "learning_rate": 9.64824120603015e-06, | |
| "loss": 2.0887, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 2.685432793807178, | |
| "grad_norm": 1.5531530380249023, | |
| "learning_rate": 9.447236180904523e-06, | |
| "loss": 1.8629, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 2.6882477128782547, | |
| "grad_norm": 1.753846287727356, | |
| "learning_rate": 9.246231155778894e-06, | |
| "loss": 1.8199, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.6910626319493316, | |
| "grad_norm": 1.5960685014724731, | |
| "learning_rate": 9.045226130653267e-06, | |
| "loss": 2.0348, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 2.693877551020408, | |
| "grad_norm": 1.6763644218444824, | |
| "learning_rate": 8.844221105527639e-06, | |
| "loss": 1.8563, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 2.696692470091485, | |
| "grad_norm": 1.4198554754257202, | |
| "learning_rate": 8.64321608040201e-06, | |
| "loss": 1.7267, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 2.6995073891625614, | |
| "grad_norm": 1.6178436279296875, | |
| "learning_rate": 8.442211055276381e-06, | |
| "loss": 2.0219, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 2.7023223082336383, | |
| "grad_norm": 1.5461405515670776, | |
| "learning_rate": 8.241206030150754e-06, | |
| "loss": 1.8272, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.705137227304715, | |
| "grad_norm": 1.7036101818084717, | |
| "learning_rate": 8.040201005025125e-06, | |
| "loss": 1.9612, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 2.7079521463757916, | |
| "grad_norm": 1.540170669555664, | |
| "learning_rate": 7.839195979899498e-06, | |
| "loss": 1.8085, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 2.7107670654468685, | |
| "grad_norm": 1.4888712167739868, | |
| "learning_rate": 7.63819095477387e-06, | |
| "loss": 1.9548, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 2.713581984517945, | |
| "grad_norm": 1.6294909715652466, | |
| "learning_rate": 7.437185929648241e-06, | |
| "loss": 1.9253, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 2.716396903589022, | |
| "grad_norm": 1.50801682472229, | |
| "learning_rate": 7.236180904522613e-06, | |
| "loss": 2.0657, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.7192118226600988, | |
| "grad_norm": 1.7348463535308838, | |
| "learning_rate": 7.035175879396985e-06, | |
| "loss": 1.9505, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 2.722026741731175, | |
| "grad_norm": 1.63502836227417, | |
| "learning_rate": 6.834170854271357e-06, | |
| "loss": 1.9629, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 2.724841660802252, | |
| "grad_norm": 1.6600550413131714, | |
| "learning_rate": 6.633165829145729e-06, | |
| "loss": 1.8303, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 2.7276565798733285, | |
| "grad_norm": 2.031532049179077, | |
| "learning_rate": 6.4321608040201015e-06, | |
| "loss": 1.7822, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 2.7304714989444054, | |
| "grad_norm": 1.804951548576355, | |
| "learning_rate": 6.231155778894473e-06, | |
| "loss": 2.1692, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.733286418015482, | |
| "grad_norm": 1.4661808013916016, | |
| "learning_rate": 6.030150753768844e-06, | |
| "loss": 1.964, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 2.736101337086559, | |
| "grad_norm": 2.2973814010620117, | |
| "learning_rate": 5.829145728643216e-06, | |
| "loss": 1.7313, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.7389162561576352, | |
| "grad_norm": 1.6403539180755615, | |
| "learning_rate": 5.628140703517588e-06, | |
| "loss": 1.9302, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 2.741731175228712, | |
| "grad_norm": 1.5790972709655762, | |
| "learning_rate": 5.4271356783919595e-06, | |
| "loss": 1.6091, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 2.744546094299789, | |
| "grad_norm": 1.765395998954773, | |
| "learning_rate": 5.226130653266332e-06, | |
| "loss": 2.0181, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.7473610133708655, | |
| "grad_norm": 1.4669338464736938, | |
| "learning_rate": 5.025125628140704e-06, | |
| "loss": 1.8627, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 2.7501759324419424, | |
| "grad_norm": 1.678789734840393, | |
| "learning_rate": 4.824120603015075e-06, | |
| "loss": 2.0305, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 2.752990851513019, | |
| "grad_norm": 1.5381121635437012, | |
| "learning_rate": 4.623115577889447e-06, | |
| "loss": 1.8813, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 2.7558057705840957, | |
| "grad_norm": 1.5611159801483154, | |
| "learning_rate": 4.422110552763819e-06, | |
| "loss": 1.7127, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 2.7586206896551726, | |
| "grad_norm": 1.6193746328353882, | |
| "learning_rate": 4.2211055276381906e-06, | |
| "loss": 1.9128, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.761435608726249, | |
| "grad_norm": 1.4393192529678345, | |
| "learning_rate": 4.020100502512563e-06, | |
| "loss": 1.5317, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 2.764250527797326, | |
| "grad_norm": 1.6281440258026123, | |
| "learning_rate": 3.819095477386935e-06, | |
| "loss": 1.9902, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 2.7670654468684024, | |
| "grad_norm": 1.6578021049499512, | |
| "learning_rate": 3.6180904522613065e-06, | |
| "loss": 2.2476, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 2.7698803659394793, | |
| "grad_norm": 1.6723147630691528, | |
| "learning_rate": 3.4170854271356786e-06, | |
| "loss": 1.9034, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 2.772695285010556, | |
| "grad_norm": 1.6200298070907593, | |
| "learning_rate": 3.2160804020100507e-06, | |
| "loss": 2.129, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.7755102040816326, | |
| "grad_norm": 1.6610552072525024, | |
| "learning_rate": 3.015075376884422e-06, | |
| "loss": 1.9335, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 2.7783251231527095, | |
| "grad_norm": 1.7721863985061646, | |
| "learning_rate": 2.814070351758794e-06, | |
| "loss": 1.888, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 2.781140042223786, | |
| "grad_norm": 1.949180006980896, | |
| "learning_rate": 2.613065326633166e-06, | |
| "loss": 2.0525, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 2.783954961294863, | |
| "grad_norm": 1.5390926599502563, | |
| "learning_rate": 2.4120603015075375e-06, | |
| "loss": 1.981, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 2.7867698803659398, | |
| "grad_norm": 1.7333327531814575, | |
| "learning_rate": 2.2110552763819096e-06, | |
| "loss": 1.7485, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.789584799437016, | |
| "grad_norm": 1.438262939453125, | |
| "learning_rate": 2.0100502512562813e-06, | |
| "loss": 1.8643, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 2.7923997185080927, | |
| "grad_norm": 1.7008702754974365, | |
| "learning_rate": 1.8090452261306533e-06, | |
| "loss": 2.0458, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 2.7952146375791695, | |
| "grad_norm": 1.8295824527740479, | |
| "learning_rate": 1.6080402010050254e-06, | |
| "loss": 2.0128, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 2.7980295566502464, | |
| "grad_norm": 1.8252149820327759, | |
| "learning_rate": 1.407035175879397e-06, | |
| "loss": 1.724, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 2.800844475721323, | |
| "grad_norm": 1.612557291984558, | |
| "learning_rate": 1.2060301507537688e-06, | |
| "loss": 1.9216, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.8036593947924, | |
| "grad_norm": 1.486989974975586, | |
| "learning_rate": 1.0050251256281407e-06, | |
| "loss": 1.6633, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 2.8064743138634762, | |
| "grad_norm": 1.5488345623016357, | |
| "learning_rate": 8.040201005025127e-07, | |
| "loss": 1.8513, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 2.809289232934553, | |
| "grad_norm": 1.741253137588501, | |
| "learning_rate": 6.030150753768844e-07, | |
| "loss": 1.7444, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 2.81210415200563, | |
| "grad_norm": 1.6252341270446777, | |
| "learning_rate": 4.0201005025125634e-07, | |
| "loss": 1.9627, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 2.8149190710767065, | |
| "grad_norm": 1.5533764362335205, | |
| "learning_rate": 2.0100502512562817e-07, | |
| "loss": 1.8062, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.447791767273472e+16, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |