| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.44483985765124556, | |
| "eval_steps": 500, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00044483985765124553, | |
| "grad_norm": 0.5906195640563965, | |
| "learning_rate": 4e-05, | |
| "loss": 1.504, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0008896797153024911, | |
| "grad_norm": 0.8538110852241516, | |
| "learning_rate": 8e-05, | |
| "loss": 1.4735, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0013345195729537367, | |
| "grad_norm": 0.4623548686504364, | |
| "learning_rate": 0.00012, | |
| "loss": 1.3887, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0017793594306049821, | |
| "grad_norm": 0.40841907262802124, | |
| "learning_rate": 0.00016, | |
| "loss": 1.2993, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.002224199288256228, | |
| "grad_norm": 0.4153430759906769, | |
| "learning_rate": 0.0002, | |
| "loss": 1.3105, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0026690391459074734, | |
| "grad_norm": 0.4014444947242737, | |
| "learning_rate": 0.00019979899497487438, | |
| "loss": 1.1942, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.003113879003558719, | |
| "grad_norm": 0.39270123839378357, | |
| "learning_rate": 0.00019959798994974876, | |
| "loss": 1.1688, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0035587188612099642, | |
| "grad_norm": 0.4414786696434021, | |
| "learning_rate": 0.00019939698492462313, | |
| "loss": 1.121, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00400355871886121, | |
| "grad_norm": 0.3911682069301605, | |
| "learning_rate": 0.0001991959798994975, | |
| "loss": 1.1039, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.004448398576512456, | |
| "grad_norm": 0.3546655476093292, | |
| "learning_rate": 0.00019899497487437187, | |
| "loss": 1.1259, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.004893238434163701, | |
| "grad_norm": 0.37283238768577576, | |
| "learning_rate": 0.00019879396984924622, | |
| "loss": 1.1599, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.005338078291814947, | |
| "grad_norm": 0.35209548473358154, | |
| "learning_rate": 0.00019859296482412062, | |
| "loss": 1.2022, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.005782918149466192, | |
| "grad_norm": 0.3735697865486145, | |
| "learning_rate": 0.000198391959798995, | |
| "loss": 1.2544, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.006227758007117438, | |
| "grad_norm": 0.324276328086853, | |
| "learning_rate": 0.00019819095477386937, | |
| "loss": 1.1451, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0066725978647686835, | |
| "grad_norm": 0.44405269622802734, | |
| "learning_rate": 0.0001979899497487437, | |
| "loss": 1.2581, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0071174377224199285, | |
| "grad_norm": 0.40987834334373474, | |
| "learning_rate": 0.0001977889447236181, | |
| "loss": 1.1422, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.007562277580071174, | |
| "grad_norm": 0.31967031955718994, | |
| "learning_rate": 0.00019758793969849249, | |
| "loss": 1.1521, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.00800711743772242, | |
| "grad_norm": 0.3634157180786133, | |
| "learning_rate": 0.00019738693467336683, | |
| "loss": 1.0172, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.008451957295373666, | |
| "grad_norm": 0.37751761078834534, | |
| "learning_rate": 0.0001971859296482412, | |
| "loss": 1.1343, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.008896797153024912, | |
| "grad_norm": 0.3501513600349426, | |
| "learning_rate": 0.0001969849246231156, | |
| "loss": 1.1137, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.009341637010676156, | |
| "grad_norm": 0.41018760204315186, | |
| "learning_rate": 0.00019678391959798995, | |
| "loss": 1.0601, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.009786476868327402, | |
| "grad_norm": 0.5334712266921997, | |
| "learning_rate": 0.00019658291457286432, | |
| "loss": 1.2332, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.010231316725978648, | |
| "grad_norm": 0.5504945516586304, | |
| "learning_rate": 0.0001963819095477387, | |
| "loss": 1.1928, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.010676156583629894, | |
| "grad_norm": 0.39751070737838745, | |
| "learning_rate": 0.0001961809045226131, | |
| "loss": 0.9774, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01112099644128114, | |
| "grad_norm": 0.31996771693229675, | |
| "learning_rate": 0.00019597989949748744, | |
| "loss": 1.042, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.011565836298932384, | |
| "grad_norm": 0.49899256229400635, | |
| "learning_rate": 0.00019577889447236181, | |
| "loss": 1.0384, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01201067615658363, | |
| "grad_norm": 0.37340229749679565, | |
| "learning_rate": 0.0001955778894472362, | |
| "loss": 1.0307, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.012455516014234875, | |
| "grad_norm": 0.33216455578804016, | |
| "learning_rate": 0.00019537688442211056, | |
| "loss": 1.0253, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.012900355871886121, | |
| "grad_norm": 0.3582470715045929, | |
| "learning_rate": 0.00019517587939698493, | |
| "loss": 1.0751, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.013345195729537367, | |
| "grad_norm": 0.3682143986225128, | |
| "learning_rate": 0.0001949748743718593, | |
| "loss": 1.0633, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.013790035587188613, | |
| "grad_norm": 0.4105437695980072, | |
| "learning_rate": 0.00019477386934673368, | |
| "loss": 1.1865, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.014234875444839857, | |
| "grad_norm": 0.3709512948989868, | |
| "learning_rate": 0.00019457286432160805, | |
| "loss": 1.0198, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.014679715302491103, | |
| "grad_norm": 0.3693946599960327, | |
| "learning_rate": 0.00019437185929648243, | |
| "loss": 1.0405, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.015124555160142349, | |
| "grad_norm": 0.335426390171051, | |
| "learning_rate": 0.0001941708542713568, | |
| "loss": 0.9301, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.015569395017793594, | |
| "grad_norm": 0.5277281403541565, | |
| "learning_rate": 0.00019396984924623117, | |
| "loss": 0.9615, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.01601423487544484, | |
| "grad_norm": 0.40925148129463196, | |
| "learning_rate": 0.00019376884422110552, | |
| "loss": 1.1093, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.016459074733096084, | |
| "grad_norm": 0.40884852409362793, | |
| "learning_rate": 0.00019356783919597992, | |
| "loss": 1.063, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.016903914590747332, | |
| "grad_norm": 0.3492753207683563, | |
| "learning_rate": 0.0001933668341708543, | |
| "loss": 1.0865, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.017348754448398576, | |
| "grad_norm": 0.3660659193992615, | |
| "learning_rate": 0.00019316582914572864, | |
| "loss": 1.071, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.017793594306049824, | |
| "grad_norm": 0.3863303065299988, | |
| "learning_rate": 0.000192964824120603, | |
| "loss": 1.0096, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.018238434163701068, | |
| "grad_norm": 0.34570327401161194, | |
| "learning_rate": 0.0001927638190954774, | |
| "loss": 1.0753, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.018683274021352312, | |
| "grad_norm": 0.35912126302719116, | |
| "learning_rate": 0.00019256281407035178, | |
| "loss": 1.0997, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.01912811387900356, | |
| "grad_norm": 0.36907467246055603, | |
| "learning_rate": 0.00019236180904522613, | |
| "loss": 1.0419, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.019572953736654804, | |
| "grad_norm": 0.3746166229248047, | |
| "learning_rate": 0.0001921608040201005, | |
| "loss": 1.0424, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.02001779359430605, | |
| "grad_norm": 0.33465832471847534, | |
| "learning_rate": 0.0001919597989949749, | |
| "loss": 1.0335, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.020462633451957295, | |
| "grad_norm": 0.3497348427772522, | |
| "learning_rate": 0.00019175879396984925, | |
| "loss": 1.0282, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02090747330960854, | |
| "grad_norm": 0.34102576971054077, | |
| "learning_rate": 0.00019155778894472362, | |
| "loss": 1.1197, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.021352313167259787, | |
| "grad_norm": 0.31394055485725403, | |
| "learning_rate": 0.000191356783919598, | |
| "loss": 1.0298, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.02179715302491103, | |
| "grad_norm": 0.505631148815155, | |
| "learning_rate": 0.0001911557788944724, | |
| "loss": 1.0349, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.02224199288256228, | |
| "grad_norm": 0.3988608121871948, | |
| "learning_rate": 0.00019095477386934674, | |
| "loss": 0.9868, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.022686832740213523, | |
| "grad_norm": 0.3646712899208069, | |
| "learning_rate": 0.0001907537688442211, | |
| "loss": 1.0884, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.023131672597864767, | |
| "grad_norm": 0.35094475746154785, | |
| "learning_rate": 0.00019055276381909548, | |
| "loss": 1.1207, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.023576512455516015, | |
| "grad_norm": 0.34856799244880676, | |
| "learning_rate": 0.00019035175879396986, | |
| "loss": 1.0121, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.02402135231316726, | |
| "grad_norm": 0.37425950169563293, | |
| "learning_rate": 0.00019015075376884423, | |
| "loss": 1.1073, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.024466192170818506, | |
| "grad_norm": 0.3625994026660919, | |
| "learning_rate": 0.0001899497487437186, | |
| "loss": 1.0863, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.02491103202846975, | |
| "grad_norm": 0.295337051153183, | |
| "learning_rate": 0.00018974874371859298, | |
| "loss": 1.0561, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.025355871886120998, | |
| "grad_norm": 0.34300458431243896, | |
| "learning_rate": 0.00018954773869346732, | |
| "loss": 0.9955, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.025800711743772242, | |
| "grad_norm": 0.45535075664520264, | |
| "learning_rate": 0.00018934673366834172, | |
| "loss": 1.0623, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.026245551601423486, | |
| "grad_norm": 0.3560766577720642, | |
| "learning_rate": 0.0001891457286432161, | |
| "loss": 1.0259, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.026690391459074734, | |
| "grad_norm": 0.33135348558425903, | |
| "learning_rate": 0.00018894472361809047, | |
| "loss": 1.0823, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.027135231316725978, | |
| "grad_norm": 0.33519864082336426, | |
| "learning_rate": 0.00018874371859296481, | |
| "loss": 0.8893, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.027580071174377226, | |
| "grad_norm": 0.4156452417373657, | |
| "learning_rate": 0.00018854271356783921, | |
| "loss": 0.888, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.02802491103202847, | |
| "grad_norm": 0.5566922426223755, | |
| "learning_rate": 0.0001883417085427136, | |
| "loss": 0.9465, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.028469750889679714, | |
| "grad_norm": 0.3567597270011902, | |
| "learning_rate": 0.00018814070351758793, | |
| "loss": 1.044, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.02891459074733096, | |
| "grad_norm": 0.3937987685203552, | |
| "learning_rate": 0.0001879396984924623, | |
| "loss": 1.1491, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.029359430604982206, | |
| "grad_norm": 0.3297279477119446, | |
| "learning_rate": 0.0001877386934673367, | |
| "loss": 1.1387, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.029804270462633453, | |
| "grad_norm": 0.4475502371788025, | |
| "learning_rate": 0.00018753768844221108, | |
| "loss": 1.1832, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.030249110320284697, | |
| "grad_norm": 0.4012911021709442, | |
| "learning_rate": 0.00018733668341708543, | |
| "loss": 1.1017, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.03069395017793594, | |
| "grad_norm": 0.36798158288002014, | |
| "learning_rate": 0.0001871356783919598, | |
| "loss": 0.9889, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.03113879003558719, | |
| "grad_norm": 0.3330882787704468, | |
| "learning_rate": 0.0001869346733668342, | |
| "loss": 1.0825, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03158362989323844, | |
| "grad_norm": 0.41344648599624634, | |
| "learning_rate": 0.00018673366834170854, | |
| "loss": 1.0599, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.03202846975088968, | |
| "grad_norm": 0.5002074837684631, | |
| "learning_rate": 0.00018653266331658292, | |
| "loss": 1.0586, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.032473309608540925, | |
| "grad_norm": 0.3306974470615387, | |
| "learning_rate": 0.0001863316582914573, | |
| "loss": 1.0004, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.03291814946619217, | |
| "grad_norm": 0.394704133272171, | |
| "learning_rate": 0.0001861306532663317, | |
| "loss": 1.0083, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.03336298932384341, | |
| "grad_norm": 0.3663751184940338, | |
| "learning_rate": 0.00018592964824120604, | |
| "loss": 0.8553, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.033807829181494664, | |
| "grad_norm": 0.37072721123695374, | |
| "learning_rate": 0.0001857286432160804, | |
| "loss": 1.1199, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.03425266903914591, | |
| "grad_norm": 0.37048038840293884, | |
| "learning_rate": 0.00018552763819095478, | |
| "loss": 1.0716, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.03469750889679715, | |
| "grad_norm": 0.34426698088645935, | |
| "learning_rate": 0.00018532663316582915, | |
| "loss": 1.0378, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.035142348754448396, | |
| "grad_norm": 0.3649255037307739, | |
| "learning_rate": 0.00018512562814070353, | |
| "loss": 1.0843, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.03558718861209965, | |
| "grad_norm": 0.37399303913116455, | |
| "learning_rate": 0.0001849246231155779, | |
| "loss": 0.9741, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03603202846975089, | |
| "grad_norm": 0.3313739001750946, | |
| "learning_rate": 0.00018472361809045227, | |
| "loss": 0.9883, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.036476868327402136, | |
| "grad_norm": 0.36365818977355957, | |
| "learning_rate": 0.00018452261306532662, | |
| "loss": 1.0434, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.03692170818505338, | |
| "grad_norm": 0.33248934149742126, | |
| "learning_rate": 0.00018432160804020102, | |
| "loss": 1.0635, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.037366548042704624, | |
| "grad_norm": 0.361134797334671, | |
| "learning_rate": 0.0001841206030150754, | |
| "loss": 1.0248, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.037811387900355875, | |
| "grad_norm": 0.3499568700790405, | |
| "learning_rate": 0.00018391959798994977, | |
| "loss": 0.9905, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.03825622775800712, | |
| "grad_norm": 0.3246215879917145, | |
| "learning_rate": 0.0001837185929648241, | |
| "loss": 1.0431, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.03870106761565836, | |
| "grad_norm": 0.39000576734542847, | |
| "learning_rate": 0.0001835175879396985, | |
| "loss": 0.995, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.03914590747330961, | |
| "grad_norm": 0.32663071155548096, | |
| "learning_rate": 0.00018331658291457288, | |
| "loss": 1.0976, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.03959074733096085, | |
| "grad_norm": 0.35704660415649414, | |
| "learning_rate": 0.00018311557788944723, | |
| "loss": 1.088, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0400355871886121, | |
| "grad_norm": 0.3753162622451782, | |
| "learning_rate": 0.0001829145728643216, | |
| "loss": 1.0509, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.04048042704626335, | |
| "grad_norm": 0.45394080877304077, | |
| "learning_rate": 0.000182713567839196, | |
| "loss": 0.8953, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.04092526690391459, | |
| "grad_norm": 0.35750091075897217, | |
| "learning_rate": 0.00018251256281407038, | |
| "loss": 1.0031, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.041370106761565835, | |
| "grad_norm": 0.2915020287036896, | |
| "learning_rate": 0.00018231155778894472, | |
| "loss": 0.9292, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.04181494661921708, | |
| "grad_norm": 0.29004380106925964, | |
| "learning_rate": 0.0001821105527638191, | |
| "loss": 0.9423, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.04225978647686833, | |
| "grad_norm": 0.3510340750217438, | |
| "learning_rate": 0.0001819095477386935, | |
| "loss": 0.9933, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.042704626334519574, | |
| "grad_norm": 0.3953409194946289, | |
| "learning_rate": 0.00018170854271356784, | |
| "loss": 1.0174, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.04314946619217082, | |
| "grad_norm": 0.3693845272064209, | |
| "learning_rate": 0.00018150753768844221, | |
| "loss": 1.1309, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.04359430604982206, | |
| "grad_norm": 0.33780723810195923, | |
| "learning_rate": 0.0001813065326633166, | |
| "loss": 0.995, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.04403914590747331, | |
| "grad_norm": 0.39444178342819214, | |
| "learning_rate": 0.00018110552763819096, | |
| "loss": 1.0567, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.04448398576512456, | |
| "grad_norm": 0.3430976867675781, | |
| "learning_rate": 0.00018090452261306533, | |
| "loss": 0.956, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0449288256227758, | |
| "grad_norm": 0.31412309408187866, | |
| "learning_rate": 0.0001807035175879397, | |
| "loss": 1.1141, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.045373665480427046, | |
| "grad_norm": 0.3321763575077057, | |
| "learning_rate": 0.00018050251256281408, | |
| "loss": 1.0245, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.04581850533807829, | |
| "grad_norm": 0.3730502128601074, | |
| "learning_rate": 0.00018030150753768845, | |
| "loss": 1.0961, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.046263345195729534, | |
| "grad_norm": 0.33242374658584595, | |
| "learning_rate": 0.00018010050251256282, | |
| "loss": 0.9899, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.046708185053380785, | |
| "grad_norm": 0.3853437900543213, | |
| "learning_rate": 0.0001798994974874372, | |
| "loss": 0.9559, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.04715302491103203, | |
| "grad_norm": 0.33745744824409485, | |
| "learning_rate": 0.00017969849246231157, | |
| "loss": 1.0458, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.04759786476868327, | |
| "grad_norm": 0.3601361811161041, | |
| "learning_rate": 0.00017949748743718592, | |
| "loss": 1.0801, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.04804270462633452, | |
| "grad_norm": 0.4157007932662964, | |
| "learning_rate": 0.00017929648241206032, | |
| "loss": 0.9852, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.04848754448398576, | |
| "grad_norm": 0.2746104598045349, | |
| "learning_rate": 0.0001790954773869347, | |
| "loss": 1.0247, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.04893238434163701, | |
| "grad_norm": 0.3099765181541443, | |
| "learning_rate": 0.00017889447236180906, | |
| "loss": 0.9642, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04937722419928826, | |
| "grad_norm": 0.3376833498477936, | |
| "learning_rate": 0.0001786934673366834, | |
| "loss": 1.0203, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.0498220640569395, | |
| "grad_norm": 0.3929193913936615, | |
| "learning_rate": 0.0001784924623115578, | |
| "loss": 1.1084, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.050266903914590745, | |
| "grad_norm": 0.3145126402378082, | |
| "learning_rate": 0.00017829145728643218, | |
| "loss": 1.124, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.050711743772241996, | |
| "grad_norm": 0.3065921664237976, | |
| "learning_rate": 0.00017809045226130653, | |
| "loss": 1.0453, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.05115658362989324, | |
| "grad_norm": 0.3715677857398987, | |
| "learning_rate": 0.0001778894472361809, | |
| "loss": 0.9841, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.051601423487544484, | |
| "grad_norm": 0.36604928970336914, | |
| "learning_rate": 0.0001776884422110553, | |
| "loss": 0.9419, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.05204626334519573, | |
| "grad_norm": 0.37098243832588196, | |
| "learning_rate": 0.00017748743718592967, | |
| "loss": 1.0133, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.05249110320284697, | |
| "grad_norm": 0.3711595833301544, | |
| "learning_rate": 0.00017728643216080402, | |
| "loss": 1.0324, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.052935943060498224, | |
| "grad_norm": 0.3096163868904114, | |
| "learning_rate": 0.0001770854271356784, | |
| "loss": 1.0086, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.05338078291814947, | |
| "grad_norm": 0.4001842737197876, | |
| "learning_rate": 0.0001768844221105528, | |
| "loss": 1.026, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.05382562277580071, | |
| "grad_norm": 0.29020652174949646, | |
| "learning_rate": 0.00017668341708542714, | |
| "loss": 0.9578, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.054270462633451956, | |
| "grad_norm": 0.35916441679000854, | |
| "learning_rate": 0.0001764824120603015, | |
| "loss": 1.0502, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0547153024911032, | |
| "grad_norm": 0.36931201815605164, | |
| "learning_rate": 0.00017628140703517588, | |
| "loss": 0.9456, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.05516014234875445, | |
| "grad_norm": 0.37013575434684753, | |
| "learning_rate": 0.00017608040201005026, | |
| "loss": 1.0472, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.055604982206405695, | |
| "grad_norm": 0.3738909959793091, | |
| "learning_rate": 0.00017587939698492463, | |
| "loss": 1.0194, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.05604982206405694, | |
| "grad_norm": 0.4676179885864258, | |
| "learning_rate": 0.000175678391959799, | |
| "loss": 1.1556, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.056494661921708184, | |
| "grad_norm": 0.430128812789917, | |
| "learning_rate": 0.00017547738693467338, | |
| "loss": 1.1472, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.05693950177935943, | |
| "grad_norm": 0.3177819848060608, | |
| "learning_rate": 0.00017527638190954775, | |
| "loss": 1.0159, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.05738434163701068, | |
| "grad_norm": 0.35535189509391785, | |
| "learning_rate": 0.00017507537688442212, | |
| "loss": 1.0388, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.05782918149466192, | |
| "grad_norm": 0.41556811332702637, | |
| "learning_rate": 0.0001748743718592965, | |
| "loss": 1.0366, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05827402135231317, | |
| "grad_norm": 0.28733861446380615, | |
| "learning_rate": 0.00017467336683417087, | |
| "loss": 1.041, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.05871886120996441, | |
| "grad_norm": 0.36875680088996887, | |
| "learning_rate": 0.00017447236180904521, | |
| "loss": 1.0408, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.059163701067615655, | |
| "grad_norm": 0.30114784836769104, | |
| "learning_rate": 0.00017427135678391961, | |
| "loss": 1.0582, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.059608540925266906, | |
| "grad_norm": 0.397324800491333, | |
| "learning_rate": 0.000174070351758794, | |
| "loss": 1.0715, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.06005338078291815, | |
| "grad_norm": 0.30825501680374146, | |
| "learning_rate": 0.00017386934673366836, | |
| "loss": 0.8977, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.060498220640569395, | |
| "grad_norm": 0.34875619411468506, | |
| "learning_rate": 0.0001736683417085427, | |
| "loss": 1.0427, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.06094306049822064, | |
| "grad_norm": 0.44976702332496643, | |
| "learning_rate": 0.0001734673366834171, | |
| "loss": 1.0594, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.06138790035587188, | |
| "grad_norm": 0.4009503722190857, | |
| "learning_rate": 0.00017326633165829148, | |
| "loss": 1.0156, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.061832740213523134, | |
| "grad_norm": 0.32680168747901917, | |
| "learning_rate": 0.00017306532663316582, | |
| "loss": 1.0518, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.06227758007117438, | |
| "grad_norm": 0.3297230005264282, | |
| "learning_rate": 0.0001728643216080402, | |
| "loss": 0.982, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.06272241992882563, | |
| "grad_norm": 0.3223113417625427, | |
| "learning_rate": 0.0001726633165829146, | |
| "loss": 0.9451, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.06316725978647687, | |
| "grad_norm": 0.3400196135044098, | |
| "learning_rate": 0.00017246231155778897, | |
| "loss": 0.9247, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.06361209964412812, | |
| "grad_norm": 0.36805739998817444, | |
| "learning_rate": 0.00017226130653266332, | |
| "loss": 1.0732, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.06405693950177936, | |
| "grad_norm": 0.40036970376968384, | |
| "learning_rate": 0.0001720603015075377, | |
| "loss": 1.0966, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.0645017793594306, | |
| "grad_norm": 0.3755742311477661, | |
| "learning_rate": 0.00017185929648241206, | |
| "loss": 0.9863, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.06494661921708185, | |
| "grad_norm": 0.4415525496006012, | |
| "learning_rate": 0.00017165829145728644, | |
| "loss": 1.0933, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.0653914590747331, | |
| "grad_norm": 0.3682146966457367, | |
| "learning_rate": 0.0001714572864321608, | |
| "loss": 0.9468, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.06583629893238434, | |
| "grad_norm": 0.3670349717140198, | |
| "learning_rate": 0.00017125628140703518, | |
| "loss": 1.1367, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.06628113879003558, | |
| "grad_norm": 0.30022603273391724, | |
| "learning_rate": 0.00017105527638190955, | |
| "loss": 0.9387, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.06672597864768683, | |
| "grad_norm": 0.36275291442871094, | |
| "learning_rate": 0.00017085427135678393, | |
| "loss": 0.8882, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06717081850533808, | |
| "grad_norm": 0.4646683633327484, | |
| "learning_rate": 0.0001706532663316583, | |
| "loss": 0.8556, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.06761565836298933, | |
| "grad_norm": 0.34833550453186035, | |
| "learning_rate": 0.00017045226130653267, | |
| "loss": 1.0124, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.06806049822064057, | |
| "grad_norm": 0.34811916947364807, | |
| "learning_rate": 0.00017025125628140705, | |
| "loss": 0.8668, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.06850533807829182, | |
| "grad_norm": 0.42011362314224243, | |
| "learning_rate": 0.00017005025125628142, | |
| "loss": 0.9818, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.06895017793594306, | |
| "grad_norm": 0.34665143489837646, | |
| "learning_rate": 0.0001698492462311558, | |
| "loss": 0.9958, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.0693950177935943, | |
| "grad_norm": 0.3390451967716217, | |
| "learning_rate": 0.00016964824120603016, | |
| "loss": 0.9835, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.06983985765124555, | |
| "grad_norm": 0.3538704812526703, | |
| "learning_rate": 0.0001694472361809045, | |
| "loss": 1.102, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.07028469750889679, | |
| "grad_norm": 0.40229830145835876, | |
| "learning_rate": 0.0001692462311557789, | |
| "loss": 1.1512, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.07072953736654804, | |
| "grad_norm": 0.37556755542755127, | |
| "learning_rate": 0.00016904522613065328, | |
| "loss": 1.0165, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.0711743772241993, | |
| "grad_norm": 0.33789652585983276, | |
| "learning_rate": 0.00016884422110552766, | |
| "loss": 0.9962, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.07161921708185054, | |
| "grad_norm": 0.39385226368904114, | |
| "learning_rate": 0.000168643216080402, | |
| "loss": 1.0538, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.07206405693950178, | |
| "grad_norm": 0.39948439598083496, | |
| "learning_rate": 0.0001684422110552764, | |
| "loss": 1.0472, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.07250889679715303, | |
| "grad_norm": 0.3786976635456085, | |
| "learning_rate": 0.00016824120603015078, | |
| "loss": 1.0288, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.07295373665480427, | |
| "grad_norm": 0.3729378879070282, | |
| "learning_rate": 0.00016804020100502512, | |
| "loss": 0.9856, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.07339857651245552, | |
| "grad_norm": 0.36763879656791687, | |
| "learning_rate": 0.0001678391959798995, | |
| "loss": 0.9711, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.07384341637010676, | |
| "grad_norm": 0.38680917024612427, | |
| "learning_rate": 0.0001676381909547739, | |
| "loss": 0.8915, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.074288256227758, | |
| "grad_norm": 0.4449096620082855, | |
| "learning_rate": 0.00016743718592964827, | |
| "loss": 0.9931, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.07473309608540925, | |
| "grad_norm": 0.39082011580467224, | |
| "learning_rate": 0.0001672361809045226, | |
| "loss": 1.0084, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.07517793594306049, | |
| "grad_norm": 0.37978312373161316, | |
| "learning_rate": 0.00016703517587939699, | |
| "loss": 0.9449, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.07562277580071175, | |
| "grad_norm": 0.32479336857795715, | |
| "learning_rate": 0.00016683417085427136, | |
| "loss": 0.8941, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.076067615658363, | |
| "grad_norm": 0.35645660758018494, | |
| "learning_rate": 0.00016663316582914573, | |
| "loss": 0.9952, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.07651245551601424, | |
| "grad_norm": 0.40985000133514404, | |
| "learning_rate": 0.0001664321608040201, | |
| "loss": 1.0187, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.07695729537366548, | |
| "grad_norm": 0.4032020568847656, | |
| "learning_rate": 0.00016623115577889448, | |
| "loss": 0.9356, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.07740213523131673, | |
| "grad_norm": 0.3428996801376343, | |
| "learning_rate": 0.00016603015075376885, | |
| "loss": 0.9326, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.07784697508896797, | |
| "grad_norm": 0.2806510925292969, | |
| "learning_rate": 0.00016582914572864322, | |
| "loss": 0.6118, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.07829181494661921, | |
| "grad_norm": 0.3010331392288208, | |
| "learning_rate": 0.0001656281407035176, | |
| "loss": 1.0065, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.07873665480427046, | |
| "grad_norm": 0.2989744246006012, | |
| "learning_rate": 0.00016542713567839197, | |
| "loss": 0.948, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.0791814946619217, | |
| "grad_norm": 0.3374849259853363, | |
| "learning_rate": 0.00016522613065326634, | |
| "loss": 1.0899, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.07962633451957295, | |
| "grad_norm": 0.38859498500823975, | |
| "learning_rate": 0.00016502512562814072, | |
| "loss": 1.035, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.0800711743772242, | |
| "grad_norm": 0.34474217891693115, | |
| "learning_rate": 0.0001648241206030151, | |
| "loss": 0.9826, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.08051601423487545, | |
| "grad_norm": 0.2957572937011719, | |
| "learning_rate": 0.00016462311557788946, | |
| "loss": 0.8905, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.0809608540925267, | |
| "grad_norm": 0.3576017916202545, | |
| "learning_rate": 0.0001644221105527638, | |
| "loss": 1.0266, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.08140569395017794, | |
| "grad_norm": 0.38848915696144104, | |
| "learning_rate": 0.0001642211055276382, | |
| "loss": 0.9883, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.08185053380782918, | |
| "grad_norm": 0.4087153375148773, | |
| "learning_rate": 0.00016402010050251258, | |
| "loss": 1.1882, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.08229537366548043, | |
| "grad_norm": 0.3468906879425049, | |
| "learning_rate": 0.00016381909547738695, | |
| "loss": 0.9818, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.08274021352313167, | |
| "grad_norm": 0.4617744982242584, | |
| "learning_rate": 0.0001636180904522613, | |
| "loss": 0.9065, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.08318505338078291, | |
| "grad_norm": 0.4207713305950165, | |
| "learning_rate": 0.0001634170854271357, | |
| "loss": 0.9988, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.08362989323843416, | |
| "grad_norm": 0.3632107675075531, | |
| "learning_rate": 0.00016321608040201007, | |
| "loss": 0.9049, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.08407473309608542, | |
| "grad_norm": 0.4303954839706421, | |
| "learning_rate": 0.00016301507537688442, | |
| "loss": 0.9801, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.08451957295373666, | |
| "grad_norm": 0.3233638107776642, | |
| "learning_rate": 0.0001628140703517588, | |
| "loss": 1.1457, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.0849644128113879, | |
| "grad_norm": 0.36270812153816223, | |
| "learning_rate": 0.00016261306532663316, | |
| "loss": 0.9804, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.08540925266903915, | |
| "grad_norm": 0.3483213484287262, | |
| "learning_rate": 0.00016241206030150756, | |
| "loss": 0.9412, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.08585409252669039, | |
| "grad_norm": 0.33415859937667847, | |
| "learning_rate": 0.0001622110552763819, | |
| "loss": 1.0479, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.08629893238434164, | |
| "grad_norm": 0.3390806317329407, | |
| "learning_rate": 0.00016201005025125628, | |
| "loss": 1.0266, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.08674377224199288, | |
| "grad_norm": 0.3439917266368866, | |
| "learning_rate": 0.00016180904522613066, | |
| "loss": 0.9495, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.08718861209964412, | |
| "grad_norm": 0.5404065847396851, | |
| "learning_rate": 0.00016160804020100503, | |
| "loss": 1.0176, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.08763345195729537, | |
| "grad_norm": 0.40834537148475647, | |
| "learning_rate": 0.0001614070351758794, | |
| "loss": 1.0119, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.08807829181494661, | |
| "grad_norm": 0.41839244961738586, | |
| "learning_rate": 0.00016120603015075378, | |
| "loss": 1.0006, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.08852313167259787, | |
| "grad_norm": 0.36179983615875244, | |
| "learning_rate": 0.00016100502512562815, | |
| "loss": 1.0247, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.08896797153024912, | |
| "grad_norm": 0.34288227558135986, | |
| "learning_rate": 0.00016080402010050252, | |
| "loss": 0.9545, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08941281138790036, | |
| "grad_norm": 0.37949255108833313, | |
| "learning_rate": 0.0001606030150753769, | |
| "loss": 1.0024, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.0898576512455516, | |
| "grad_norm": 0.4100090563297272, | |
| "learning_rate": 0.00016040201005025127, | |
| "loss": 1.0465, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.09030249110320285, | |
| "grad_norm": 0.3741397559642792, | |
| "learning_rate": 0.00016020100502512564, | |
| "loss": 0.8969, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.09074733096085409, | |
| "grad_norm": 0.4092821180820465, | |
| "learning_rate": 0.00016, | |
| "loss": 1.3054, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.09119217081850534, | |
| "grad_norm": 0.29491907358169556, | |
| "learning_rate": 0.00015979899497487439, | |
| "loss": 1.0039, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.09163701067615658, | |
| "grad_norm": 0.38052767515182495, | |
| "learning_rate": 0.00015959798994974876, | |
| "loss": 0.9034, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.09208185053380782, | |
| "grad_norm": 0.34436798095703125, | |
| "learning_rate": 0.0001593969849246231, | |
| "loss": 0.9621, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.09252669039145907, | |
| "grad_norm": 0.41748183965682983, | |
| "learning_rate": 0.0001591959798994975, | |
| "loss": 1.0112, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.09297153024911033, | |
| "grad_norm": 0.35025399923324585, | |
| "learning_rate": 0.00015899497487437188, | |
| "loss": 1.0594, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.09341637010676157, | |
| "grad_norm": 0.3578777611255646, | |
| "learning_rate": 0.00015879396984924625, | |
| "loss": 1.0191, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.09386120996441281, | |
| "grad_norm": 0.3903699517250061, | |
| "learning_rate": 0.0001585929648241206, | |
| "loss": 1.0492, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.09430604982206406, | |
| "grad_norm": 0.34571701288223267, | |
| "learning_rate": 0.000158391959798995, | |
| "loss": 1.07, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.0947508896797153, | |
| "grad_norm": 0.3766401410102844, | |
| "learning_rate": 0.00015819095477386937, | |
| "loss": 0.9508, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.09519572953736655, | |
| "grad_norm": 0.3765754699707031, | |
| "learning_rate": 0.00015798994974874372, | |
| "loss": 0.9604, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.09564056939501779, | |
| "grad_norm": 0.39237186312675476, | |
| "learning_rate": 0.0001577889447236181, | |
| "loss": 0.9678, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.09608540925266904, | |
| "grad_norm": 0.43379107117652893, | |
| "learning_rate": 0.00015758793969849246, | |
| "loss": 0.9569, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.09653024911032028, | |
| "grad_norm": 0.44235309958457947, | |
| "learning_rate": 0.00015738693467336686, | |
| "loss": 1.047, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.09697508896797152, | |
| "grad_norm": 0.37492284178733826, | |
| "learning_rate": 0.0001571859296482412, | |
| "loss": 0.9247, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.09741992882562278, | |
| "grad_norm": 0.44187766313552856, | |
| "learning_rate": 0.00015698492462311558, | |
| "loss": 1.0136, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.09786476868327403, | |
| "grad_norm": 0.41393113136291504, | |
| "learning_rate": 0.00015678391959798995, | |
| "loss": 1.0051, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.09830960854092527, | |
| "grad_norm": 0.3643686771392822, | |
| "learning_rate": 0.00015658291457286433, | |
| "loss": 1.1171, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.09875444839857651, | |
| "grad_norm": 0.351253479719162, | |
| "learning_rate": 0.0001563819095477387, | |
| "loss": 1.0466, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.09919928825622776, | |
| "grad_norm": 0.35708218812942505, | |
| "learning_rate": 0.00015618090452261307, | |
| "loss": 1.0688, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.099644128113879, | |
| "grad_norm": 0.3984560966491699, | |
| "learning_rate": 0.00015597989949748745, | |
| "loss": 0.9606, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.10008896797153025, | |
| "grad_norm": 0.3412669897079468, | |
| "learning_rate": 0.00015577889447236182, | |
| "loss": 0.9781, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.10053380782918149, | |
| "grad_norm": 0.36613014340400696, | |
| "learning_rate": 0.0001555778894472362, | |
| "loss": 0.9449, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.10097864768683273, | |
| "grad_norm": 0.3568494915962219, | |
| "learning_rate": 0.00015537688442211056, | |
| "loss": 0.9679, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.10142348754448399, | |
| "grad_norm": 0.4687019884586334, | |
| "learning_rate": 0.00015517587939698494, | |
| "loss": 1.0386, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.10186832740213524, | |
| "grad_norm": 0.40447989106178284, | |
| "learning_rate": 0.0001549748743718593, | |
| "loss": 0.9812, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.10231316725978648, | |
| "grad_norm": 0.39272576570510864, | |
| "learning_rate": 0.00015477386934673368, | |
| "loss": 1.0165, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.10275800711743772, | |
| "grad_norm": 0.3780848979949951, | |
| "learning_rate": 0.00015457286432160806, | |
| "loss": 0.9606, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.10320284697508897, | |
| "grad_norm": 0.46794167160987854, | |
| "learning_rate": 0.0001543718592964824, | |
| "loss": 0.9305, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.10364768683274021, | |
| "grad_norm": 0.39729002118110657, | |
| "learning_rate": 0.0001541708542713568, | |
| "loss": 1.0113, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.10409252669039146, | |
| "grad_norm": 0.4045659005641937, | |
| "learning_rate": 0.00015396984924623117, | |
| "loss": 0.9961, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.1045373665480427, | |
| "grad_norm": 0.35122302174568176, | |
| "learning_rate": 0.00015376884422110555, | |
| "loss": 0.9891, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.10498220640569395, | |
| "grad_norm": 0.37050357460975647, | |
| "learning_rate": 0.0001535678391959799, | |
| "loss": 1.0024, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.10542704626334519, | |
| "grad_norm": 0.40677326917648315, | |
| "learning_rate": 0.00015336683417085427, | |
| "loss": 1.0948, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.10587188612099645, | |
| "grad_norm": 0.35067057609558105, | |
| "learning_rate": 0.00015316582914572867, | |
| "loss": 0.9897, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.10631672597864769, | |
| "grad_norm": 0.3920201063156128, | |
| "learning_rate": 0.000152964824120603, | |
| "loss": 1.0114, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.10676156583629894, | |
| "grad_norm": 0.3613733649253845, | |
| "learning_rate": 0.00015276381909547739, | |
| "loss": 0.9508, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.10720640569395018, | |
| "grad_norm": 0.35030075907707214, | |
| "learning_rate": 0.00015256281407035176, | |
| "loss": 1.0019, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.10765124555160142, | |
| "grad_norm": 0.4036657512187958, | |
| "learning_rate": 0.00015236180904522613, | |
| "loss": 0.9347, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.10809608540925267, | |
| "grad_norm": 0.3680381178855896, | |
| "learning_rate": 0.0001521608040201005, | |
| "loss": 0.9541, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.10854092526690391, | |
| "grad_norm": 0.7577487826347351, | |
| "learning_rate": 0.00015195979899497488, | |
| "loss": 1.0019, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.10898576512455516, | |
| "grad_norm": 0.32293733954429626, | |
| "learning_rate": 0.00015175879396984925, | |
| "loss": 0.9768, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.1094306049822064, | |
| "grad_norm": 0.32959771156311035, | |
| "learning_rate": 0.00015155778894472362, | |
| "loss": 1.0606, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.10987544483985764, | |
| "grad_norm": 0.3493881821632385, | |
| "learning_rate": 0.000151356783919598, | |
| "loss": 1.0317, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.1103202846975089, | |
| "grad_norm": 0.3482673168182373, | |
| "learning_rate": 0.00015115577889447237, | |
| "loss": 0.9157, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.11076512455516015, | |
| "grad_norm": 0.35461658239364624, | |
| "learning_rate": 0.00015095477386934674, | |
| "loss": 1.0753, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.11120996441281139, | |
| "grad_norm": 0.3124016225337982, | |
| "learning_rate": 0.00015075376884422112, | |
| "loss": 0.9352, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.11165480427046263, | |
| "grad_norm": 0.3451753854751587, | |
| "learning_rate": 0.0001505527638190955, | |
| "loss": 0.948, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.11209964412811388, | |
| "grad_norm": 0.3436938524246216, | |
| "learning_rate": 0.00015035175879396986, | |
| "loss": 0.9277, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.11254448398576512, | |
| "grad_norm": 0.3544940650463104, | |
| "learning_rate": 0.00015015075376884423, | |
| "loss": 0.976, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.11298932384341637, | |
| "grad_norm": 0.34947720170021057, | |
| "learning_rate": 0.0001499497487437186, | |
| "loss": 1.0602, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.11343416370106761, | |
| "grad_norm": 0.38455042243003845, | |
| "learning_rate": 0.00014974874371859298, | |
| "loss": 0.9524, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.11387900355871886, | |
| "grad_norm": 0.4009455442428589, | |
| "learning_rate": 0.00014954773869346735, | |
| "loss": 1.0465, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.11432384341637011, | |
| "grad_norm": 0.5961493849754333, | |
| "learning_rate": 0.0001493467336683417, | |
| "loss": 0.8681, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.11476868327402136, | |
| "grad_norm": 0.47233107686042786, | |
| "learning_rate": 0.0001491457286432161, | |
| "loss": 0.9355, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.1152135231316726, | |
| "grad_norm": 0.41602540016174316, | |
| "learning_rate": 0.00014894472361809047, | |
| "loss": 1.0256, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.11565836298932385, | |
| "grad_norm": 0.3379324972629547, | |
| "learning_rate": 0.00014874371859296482, | |
| "loss": 0.8982, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.11610320284697509, | |
| "grad_norm": 0.3502661883831024, | |
| "learning_rate": 0.0001485427135678392, | |
| "loss": 0.8932, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.11654804270462633, | |
| "grad_norm": 0.35349032282829285, | |
| "learning_rate": 0.00014834170854271356, | |
| "loss": 0.9541, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.11699288256227758, | |
| "grad_norm": 0.37270882725715637, | |
| "learning_rate": 0.00014814070351758796, | |
| "loss": 0.9566, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.11743772241992882, | |
| "grad_norm": 0.3453938364982605, | |
| "learning_rate": 0.0001479396984924623, | |
| "loss": 0.9901, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.11788256227758007, | |
| "grad_norm": 0.4339182376861572, | |
| "learning_rate": 0.00014773869346733668, | |
| "loss": 0.9542, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.11832740213523131, | |
| "grad_norm": 0.3721541166305542, | |
| "learning_rate": 0.00014753768844221106, | |
| "loss": 0.9075, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.11877224199288257, | |
| "grad_norm": 0.3452603220939636, | |
| "learning_rate": 0.00014733668341708543, | |
| "loss": 0.9762, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.11921708185053381, | |
| "grad_norm": 0.33102792501449585, | |
| "learning_rate": 0.0001471356783919598, | |
| "loss": 0.9648, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.11966192170818506, | |
| "grad_norm": 0.38129130005836487, | |
| "learning_rate": 0.00014693467336683417, | |
| "loss": 1.0283, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.1201067615658363, | |
| "grad_norm": 0.3465835452079773, | |
| "learning_rate": 0.00014673366834170855, | |
| "loss": 0.9383, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.12055160142348754, | |
| "grad_norm": 0.31291693449020386, | |
| "learning_rate": 0.00014653266331658292, | |
| "loss": 1.0326, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.12099644128113879, | |
| "grad_norm": 0.40111976861953735, | |
| "learning_rate": 0.0001463316582914573, | |
| "loss": 1.0133, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.12144128113879003, | |
| "grad_norm": 0.3763119578361511, | |
| "learning_rate": 0.00014613065326633167, | |
| "loss": 1.0317, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.12188612099644128, | |
| "grad_norm": 0.3577355742454529, | |
| "learning_rate": 0.00014592964824120604, | |
| "loss": 1.0477, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.12233096085409252, | |
| "grad_norm": 0.3452180027961731, | |
| "learning_rate": 0.0001457286432160804, | |
| "loss": 1.0083, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.12277580071174377, | |
| "grad_norm": 0.40135836601257324, | |
| "learning_rate": 0.00014552763819095479, | |
| "loss": 0.9175, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.12322064056939502, | |
| "grad_norm": 0.45712020993232727, | |
| "learning_rate": 0.00014532663316582916, | |
| "loss": 0.9563, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.12366548042704627, | |
| "grad_norm": 0.3986268937587738, | |
| "learning_rate": 0.00014512562814070353, | |
| "loss": 0.926, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.12411032028469751, | |
| "grad_norm": 0.3132750988006592, | |
| "learning_rate": 0.0001449246231155779, | |
| "loss": 0.9503, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.12455516014234876, | |
| "grad_norm": 0.34486275911331177, | |
| "learning_rate": 0.00014472361809045228, | |
| "loss": 0.9375, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 0.3309858739376068, | |
| "learning_rate": 0.00014452261306532665, | |
| "loss": 0.9276, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.12544483985765126, | |
| "grad_norm": 0.3567025363445282, | |
| "learning_rate": 0.000144321608040201, | |
| "loss": 0.9924, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.1258896797153025, | |
| "grad_norm": 0.35258856415748596, | |
| "learning_rate": 0.00014412060301507537, | |
| "loss": 0.974, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.12633451957295375, | |
| "grad_norm": 0.33701589703559875, | |
| "learning_rate": 0.00014391959798994977, | |
| "loss": 1.0021, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.12677935943060498, | |
| "grad_norm": 0.3263789117336273, | |
| "learning_rate": 0.00014371859296482411, | |
| "loss": 1.051, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.12722419928825623, | |
| "grad_norm": 0.39523646235466003, | |
| "learning_rate": 0.0001435175879396985, | |
| "loss": 0.9349, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.12766903914590746, | |
| "grad_norm": 0.3561248183250427, | |
| "learning_rate": 0.00014331658291457286, | |
| "loss": 0.869, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.12811387900355872, | |
| "grad_norm": 0.43630316853523254, | |
| "learning_rate": 0.00014311557788944726, | |
| "loss": 0.9773, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.12855871886120995, | |
| "grad_norm": 0.3555675446987152, | |
| "learning_rate": 0.0001429145728643216, | |
| "loss": 1.0048, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.1290035587188612, | |
| "grad_norm": 0.40323683619499207, | |
| "learning_rate": 0.00014271356783919598, | |
| "loss": 0.999, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.12944839857651247, | |
| "grad_norm": 0.3344346284866333, | |
| "learning_rate": 0.00014251256281407035, | |
| "loss": 0.9882, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.1298932384341637, | |
| "grad_norm": 0.3018104135990143, | |
| "learning_rate": 0.00014231155778894473, | |
| "loss": 0.9665, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.13033807829181496, | |
| "grad_norm": 0.3336434066295624, | |
| "learning_rate": 0.0001421105527638191, | |
| "loss": 0.9786, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.1307829181494662, | |
| "grad_norm": 0.3364970088005066, | |
| "learning_rate": 0.00014190954773869347, | |
| "loss": 0.9436, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.13122775800711745, | |
| "grad_norm": 0.3738161027431488, | |
| "learning_rate": 0.00014170854271356784, | |
| "loss": 0.9884, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.13167259786476868, | |
| "grad_norm": 0.34548112750053406, | |
| "learning_rate": 0.00014150753768844222, | |
| "loss": 0.9578, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.13211743772241993, | |
| "grad_norm": 0.4063248336315155, | |
| "learning_rate": 0.0001413065326633166, | |
| "loss": 1.0649, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.13256227758007116, | |
| "grad_norm": 0.4038452208042145, | |
| "learning_rate": 0.00014110552763819096, | |
| "loss": 0.9824, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.13300711743772242, | |
| "grad_norm": 0.3010196089744568, | |
| "learning_rate": 0.00014090452261306534, | |
| "loss": 0.9492, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.13345195729537365, | |
| "grad_norm": 0.3451170325279236, | |
| "learning_rate": 0.0001407035175879397, | |
| "loss": 1.1038, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1338967971530249, | |
| "grad_norm": 0.3674055337905884, | |
| "learning_rate": 0.00014050251256281408, | |
| "loss": 0.9764, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.13434163701067617, | |
| "grad_norm": 0.28219959139823914, | |
| "learning_rate": 0.00014030150753768846, | |
| "loss": 0.8152, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.1347864768683274, | |
| "grad_norm": 0.3264709711074829, | |
| "learning_rate": 0.0001401005025125628, | |
| "loss": 1.0336, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.13523131672597866, | |
| "grad_norm": 0.3442930579185486, | |
| "learning_rate": 0.0001398994974874372, | |
| "loss": 0.9622, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.1356761565836299, | |
| "grad_norm": 0.4365461468696594, | |
| "learning_rate": 0.00013969849246231157, | |
| "loss": 0.9243, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.13612099644128114, | |
| "grad_norm": 0.3317567706108093, | |
| "learning_rate": 0.00013949748743718595, | |
| "loss": 0.9619, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.13656583629893237, | |
| "grad_norm": 0.3600742816925049, | |
| "learning_rate": 0.0001392964824120603, | |
| "loss": 1.0869, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.13701067615658363, | |
| "grad_norm": 0.414174348115921, | |
| "learning_rate": 0.00013909547738693467, | |
| "loss": 0.9696, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.13745551601423486, | |
| "grad_norm": 0.3307357430458069, | |
| "learning_rate": 0.00013889447236180907, | |
| "loss": 0.8673, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.13790035587188612, | |
| "grad_norm": 0.3574574291706085, | |
| "learning_rate": 0.0001386934673366834, | |
| "loss": 1.0225, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.13834519572953738, | |
| "grad_norm": 0.3453276455402374, | |
| "learning_rate": 0.00013849246231155778, | |
| "loss": 0.956, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.1387900355871886, | |
| "grad_norm": 0.3369787633419037, | |
| "learning_rate": 0.00013829145728643216, | |
| "loss": 0.9878, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.13923487544483987, | |
| "grad_norm": 0.36153116822242737, | |
| "learning_rate": 0.00013809045226130656, | |
| "loss": 0.9906, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.1396797153024911, | |
| "grad_norm": 0.3499455749988556, | |
| "learning_rate": 0.0001378894472361809, | |
| "loss": 1.0581, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.14012455516014236, | |
| "grad_norm": 0.3346167206764221, | |
| "learning_rate": 0.00013768844221105528, | |
| "loss": 1.1294, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.14056939501779359, | |
| "grad_norm": 0.3535563051700592, | |
| "learning_rate": 0.00013748743718592965, | |
| "loss": 1.0081, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.14101423487544484, | |
| "grad_norm": 0.3797767758369446, | |
| "learning_rate": 0.00013728643216080402, | |
| "loss": 0.9821, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.14145907473309607, | |
| "grad_norm": 0.35535991191864014, | |
| "learning_rate": 0.0001370854271356784, | |
| "loss": 0.9698, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.14190391459074733, | |
| "grad_norm": 0.3628135323524475, | |
| "learning_rate": 0.00013688442211055277, | |
| "loss": 0.8782, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.1423487544483986, | |
| "grad_norm": 0.3347620666027069, | |
| "learning_rate": 0.00013668341708542714, | |
| "loss": 0.9383, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.14279359430604982, | |
| "grad_norm": 0.3387628197669983, | |
| "learning_rate": 0.00013648241206030151, | |
| "loss": 0.9846, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.14323843416370108, | |
| "grad_norm": 0.33139505982398987, | |
| "learning_rate": 0.0001362814070351759, | |
| "loss": 0.9541, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.1436832740213523, | |
| "grad_norm": 0.3996013104915619, | |
| "learning_rate": 0.00013608040201005026, | |
| "loss": 0.9682, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.14412811387900357, | |
| "grad_norm": 0.3609457314014435, | |
| "learning_rate": 0.00013587939698492463, | |
| "loss": 0.9758, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.1445729537366548, | |
| "grad_norm": 0.31257274746894836, | |
| "learning_rate": 0.000135678391959799, | |
| "loss": 0.9793, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.14501779359430605, | |
| "grad_norm": 0.3800847828388214, | |
| "learning_rate": 0.00013547738693467338, | |
| "loss": 1.0033, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.14546263345195729, | |
| "grad_norm": 0.32256045937538147, | |
| "learning_rate": 0.00013527638190954775, | |
| "loss": 0.9944, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.14590747330960854, | |
| "grad_norm": 0.36587563157081604, | |
| "learning_rate": 0.0001350753768844221, | |
| "loss": 1.0223, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.14635231316725977, | |
| "grad_norm": 0.37276628613471985, | |
| "learning_rate": 0.00013487437185929647, | |
| "loss": 0.8559, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.14679715302491103, | |
| "grad_norm": 0.3412066698074341, | |
| "learning_rate": 0.00013467336683417087, | |
| "loss": 0.8826, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.1472419928825623, | |
| "grad_norm": 0.33895769715309143, | |
| "learning_rate": 0.00013447236180904524, | |
| "loss": 0.9517, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.14768683274021352, | |
| "grad_norm": 0.3830057680606842, | |
| "learning_rate": 0.0001342713567839196, | |
| "loss": 1.0307, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.14813167259786478, | |
| "grad_norm": 0.3927343487739563, | |
| "learning_rate": 0.00013407035175879396, | |
| "loss": 0.9916, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.148576512455516, | |
| "grad_norm": 0.3220495283603668, | |
| "learning_rate": 0.00013386934673366836, | |
| "loss": 0.9876, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.14902135231316727, | |
| "grad_norm": 0.35559162497520447, | |
| "learning_rate": 0.0001336683417085427, | |
| "loss": 1.0066, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.1494661921708185, | |
| "grad_norm": 0.36672064661979675, | |
| "learning_rate": 0.00013346733668341708, | |
| "loss": 0.9865, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.14991103202846975, | |
| "grad_norm": 0.3622550964355469, | |
| "learning_rate": 0.00013326633165829146, | |
| "loss": 1.0956, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.15035587188612098, | |
| "grad_norm": 0.3213232159614563, | |
| "learning_rate": 0.00013306532663316586, | |
| "loss": 0.9552, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.15080071174377224, | |
| "grad_norm": 0.3889192044734955, | |
| "learning_rate": 0.0001328643216080402, | |
| "loss": 0.9417, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.1512455516014235, | |
| "grad_norm": 0.32667845487594604, | |
| "learning_rate": 0.00013266331658291457, | |
| "loss": 0.9319, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.15169039145907473, | |
| "grad_norm": 0.44496315717697144, | |
| "learning_rate": 0.00013246231155778895, | |
| "loss": 0.7937, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.152135231316726, | |
| "grad_norm": 0.371467649936676, | |
| "learning_rate": 0.00013226130653266332, | |
| "loss": 0.9998, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.15258007117437722, | |
| "grad_norm": 0.3509642779827118, | |
| "learning_rate": 0.0001320603015075377, | |
| "loss": 0.9615, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.15302491103202848, | |
| "grad_norm": 0.32483533024787903, | |
| "learning_rate": 0.00013185929648241207, | |
| "loss": 0.9976, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.1534697508896797, | |
| "grad_norm": 0.38299068808555603, | |
| "learning_rate": 0.00013165829145728644, | |
| "loss": 0.8436, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.15391459074733096, | |
| "grad_norm": 0.36412617564201355, | |
| "learning_rate": 0.0001314572864321608, | |
| "loss": 1.0338, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.1543594306049822, | |
| "grad_norm": 0.3295581638813019, | |
| "learning_rate": 0.00013125628140703518, | |
| "loss": 0.9696, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.15480427046263345, | |
| "grad_norm": 0.34034737944602966, | |
| "learning_rate": 0.00013105527638190956, | |
| "loss": 0.9977, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.1552491103202847, | |
| "grad_norm": 0.3623081147670746, | |
| "learning_rate": 0.00013085427135678393, | |
| "loss": 1.0032, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.15569395017793594, | |
| "grad_norm": 0.3889707922935486, | |
| "learning_rate": 0.0001306532663316583, | |
| "loss": 0.9245, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.1561387900355872, | |
| "grad_norm": 0.3887019455432892, | |
| "learning_rate": 0.00013045226130653268, | |
| "loss": 1.0836, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.15658362989323843, | |
| "grad_norm": 0.32099735736846924, | |
| "learning_rate": 0.00013025125628140705, | |
| "loss": 0.907, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.1570284697508897, | |
| "grad_norm": 0.29804012179374695, | |
| "learning_rate": 0.0001300502512562814, | |
| "loss": 0.9691, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.15747330960854092, | |
| "grad_norm": 0.3739759922027588, | |
| "learning_rate": 0.00012984924623115577, | |
| "loss": 1.0826, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.15791814946619218, | |
| "grad_norm": 0.3806516230106354, | |
| "learning_rate": 0.00012964824120603017, | |
| "loss": 1.0638, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.1583629893238434, | |
| "grad_norm": 0.3504475951194763, | |
| "learning_rate": 0.00012944723618090454, | |
| "loss": 0.8802, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.15880782918149466, | |
| "grad_norm": 0.4315776228904724, | |
| "learning_rate": 0.0001292462311557789, | |
| "loss": 1.0159, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.1592526690391459, | |
| "grad_norm": 0.3647462725639343, | |
| "learning_rate": 0.00012904522613065326, | |
| "loss": 0.985, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.15969750889679715, | |
| "grad_norm": 0.30111998319625854, | |
| "learning_rate": 0.00012884422110552766, | |
| "loss": 0.951, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.1601423487544484, | |
| "grad_norm": 0.3357127904891968, | |
| "learning_rate": 0.000128643216080402, | |
| "loss": 1.1002, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.16058718861209964, | |
| "grad_norm": 0.4213799834251404, | |
| "learning_rate": 0.00012844221105527638, | |
| "loss": 0.8264, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.1610320284697509, | |
| "grad_norm": 0.4168066084384918, | |
| "learning_rate": 0.00012824120603015075, | |
| "loss": 1.0548, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.16147686832740213, | |
| "grad_norm": 0.36040782928466797, | |
| "learning_rate": 0.00012804020100502515, | |
| "loss": 1.0988, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.1619217081850534, | |
| "grad_norm": 0.315931499004364, | |
| "learning_rate": 0.0001278391959798995, | |
| "loss": 0.8874, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.16236654804270462, | |
| "grad_norm": 0.3526541292667389, | |
| "learning_rate": 0.00012763819095477387, | |
| "loss": 0.8408, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.16281138790035588, | |
| "grad_norm": 0.315824031829834, | |
| "learning_rate": 0.00012743718592964824, | |
| "loss": 0.9537, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.1632562277580071, | |
| "grad_norm": 0.36629655957221985, | |
| "learning_rate": 0.00012723618090452262, | |
| "loss": 0.9732, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.16370106761565836, | |
| "grad_norm": 0.3385011553764343, | |
| "learning_rate": 0.000127035175879397, | |
| "loss": 1.0207, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.16414590747330962, | |
| "grad_norm": 0.43948590755462646, | |
| "learning_rate": 0.00012683417085427136, | |
| "loss": 0.8719, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.16459074733096085, | |
| "grad_norm": 0.4001463055610657, | |
| "learning_rate": 0.00012663316582914574, | |
| "loss": 1.085, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.1650355871886121, | |
| "grad_norm": 0.37441739439964294, | |
| "learning_rate": 0.0001264321608040201, | |
| "loss": 0.9799, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.16548042704626334, | |
| "grad_norm": 0.29792410135269165, | |
| "learning_rate": 0.00012623115577889448, | |
| "loss": 0.8901, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.1659252669039146, | |
| "grad_norm": 0.42688125371932983, | |
| "learning_rate": 0.00012603015075376885, | |
| "loss": 1.085, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.16637010676156583, | |
| "grad_norm": 0.3854696750640869, | |
| "learning_rate": 0.00012582914572864323, | |
| "loss": 0.9271, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.16681494661921709, | |
| "grad_norm": 0.40170320868492126, | |
| "learning_rate": 0.0001256281407035176, | |
| "loss": 0.9355, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.16725978647686832, | |
| "grad_norm": 0.3294379711151123, | |
| "learning_rate": 0.00012542713567839197, | |
| "loss": 0.9221, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.16770462633451957, | |
| "grad_norm": 0.3526048958301544, | |
| "learning_rate": 0.00012522613065326635, | |
| "loss": 1.0483, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.16814946619217083, | |
| "grad_norm": 0.3107386529445648, | |
| "learning_rate": 0.0001250251256281407, | |
| "loss": 0.9166, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.16859430604982206, | |
| "grad_norm": 0.4228864908218384, | |
| "learning_rate": 0.00012482412060301507, | |
| "loss": 1.0207, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.16903914590747332, | |
| "grad_norm": 0.3822677731513977, | |
| "learning_rate": 0.00012462311557788947, | |
| "loss": 0.9935, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.16948398576512455, | |
| "grad_norm": 0.3852001428604126, | |
| "learning_rate": 0.00012442211055276384, | |
| "loss": 0.9831, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.1699288256227758, | |
| "grad_norm": 0.3484053313732147, | |
| "learning_rate": 0.00012422110552763818, | |
| "loss": 0.9327, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.17037366548042704, | |
| "grad_norm": 0.3499116897583008, | |
| "learning_rate": 0.00012402010050251256, | |
| "loss": 1.0521, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.1708185053380783, | |
| "grad_norm": 0.362118661403656, | |
| "learning_rate": 0.00012381909547738696, | |
| "loss": 1.0482, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.17126334519572953, | |
| "grad_norm": 0.3913699686527252, | |
| "learning_rate": 0.0001236180904522613, | |
| "loss": 0.9908, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.17170818505338079, | |
| "grad_norm": 0.3362598121166229, | |
| "learning_rate": 0.00012341708542713568, | |
| "loss": 0.9814, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.17215302491103202, | |
| "grad_norm": 0.34024205803871155, | |
| "learning_rate": 0.00012321608040201005, | |
| "loss": 0.9747, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.17259786476868327, | |
| "grad_norm": 0.35168662667274475, | |
| "learning_rate": 0.00012301507537688445, | |
| "loss": 1.0067, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.17304270462633453, | |
| "grad_norm": 0.30861225724220276, | |
| "learning_rate": 0.0001228140703517588, | |
| "loss": 1.06, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.17348754448398576, | |
| "grad_norm": 0.3759268522262573, | |
| "learning_rate": 0.00012261306532663317, | |
| "loss": 1.1586, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.17393238434163702, | |
| "grad_norm": 0.3894938826560974, | |
| "learning_rate": 0.00012241206030150754, | |
| "loss": 0.9451, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.17437722419928825, | |
| "grad_norm": 0.3796166777610779, | |
| "learning_rate": 0.00012221105527638191, | |
| "loss": 0.9602, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.1748220640569395, | |
| "grad_norm": 0.32060232758522034, | |
| "learning_rate": 0.00012201005025125629, | |
| "loss": 1.0365, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.17526690391459074, | |
| "grad_norm": 0.33367520570755005, | |
| "learning_rate": 0.00012180904522613066, | |
| "loss": 0.9616, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.175711743772242, | |
| "grad_norm": 0.3084149658679962, | |
| "learning_rate": 0.00012160804020100502, | |
| "loss": 0.8514, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.17615658362989323, | |
| "grad_norm": 0.32909664511680603, | |
| "learning_rate": 0.00012140703517587942, | |
| "loss": 0.9773, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.17660142348754448, | |
| "grad_norm": 0.3140466511249542, | |
| "learning_rate": 0.00012120603015075378, | |
| "loss": 0.9808, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.17704626334519574, | |
| "grad_norm": 0.34496647119522095, | |
| "learning_rate": 0.00012100502512562815, | |
| "loss": 0.8233, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.17749110320284697, | |
| "grad_norm": 0.4260810315608978, | |
| "learning_rate": 0.00012080402010050251, | |
| "loss": 0.9703, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.17793594306049823, | |
| "grad_norm": 0.3465157747268677, | |
| "learning_rate": 0.00012060301507537688, | |
| "loss": 1.0121, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.17838078291814946, | |
| "grad_norm": 0.3023923635482788, | |
| "learning_rate": 0.00012040201005025127, | |
| "loss": 0.9899, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.17882562277580072, | |
| "grad_norm": 0.3486076593399048, | |
| "learning_rate": 0.00012020100502512563, | |
| "loss": 0.9325, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.17927046263345195, | |
| "grad_norm": 0.3666176199913025, | |
| "learning_rate": 0.00012, | |
| "loss": 1.019, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.1797153024911032, | |
| "grad_norm": 0.3321322798728943, | |
| "learning_rate": 0.00011979899497487436, | |
| "loss": 0.9197, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.18016014234875444, | |
| "grad_norm": 0.4836527407169342, | |
| "learning_rate": 0.00011959798994974876, | |
| "loss": 0.8488, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.1806049822064057, | |
| "grad_norm": 0.3428821861743927, | |
| "learning_rate": 0.00011939698492462312, | |
| "loss": 1.0314, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.18104982206405695, | |
| "grad_norm": 0.38532841205596924, | |
| "learning_rate": 0.0001191959798994975, | |
| "loss": 0.9393, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.18149466192170818, | |
| "grad_norm": 0.3802700340747833, | |
| "learning_rate": 0.00011899497487437185, | |
| "loss": 0.9136, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.18193950177935944, | |
| "grad_norm": 0.3403851389884949, | |
| "learning_rate": 0.00011879396984924624, | |
| "loss": 0.9552, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.18238434163701067, | |
| "grad_norm": 0.3999098241329193, | |
| "learning_rate": 0.00011859296482412061, | |
| "loss": 0.9292, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.18282918149466193, | |
| "grad_norm": 0.3740805685520172, | |
| "learning_rate": 0.00011839195979899497, | |
| "loss": 1.0809, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.18327402135231316, | |
| "grad_norm": 0.42133694887161255, | |
| "learning_rate": 0.00011819095477386935, | |
| "loss": 0.8661, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.18371886120996442, | |
| "grad_norm": 0.39346155524253845, | |
| "learning_rate": 0.00011798994974874373, | |
| "loss": 1.0682, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.18416370106761565, | |
| "grad_norm": 0.3395063281059265, | |
| "learning_rate": 0.0001177889447236181, | |
| "loss": 0.9672, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.1846085409252669, | |
| "grad_norm": 0.3930981755256653, | |
| "learning_rate": 0.00011758793969849247, | |
| "loss": 0.9354, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.18505338078291814, | |
| "grad_norm": 0.40588700771331787, | |
| "learning_rate": 0.00011738693467336684, | |
| "loss": 0.9723, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.1854982206405694, | |
| "grad_norm": 0.4842967689037323, | |
| "learning_rate": 0.00011718592964824122, | |
| "loss": 0.9098, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.18594306049822065, | |
| "grad_norm": 0.34736424684524536, | |
| "learning_rate": 0.00011698492462311558, | |
| "loss": 0.9383, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.18638790035587188, | |
| "grad_norm": 0.41721245646476746, | |
| "learning_rate": 0.00011678391959798996, | |
| "loss": 0.9804, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.18683274021352314, | |
| "grad_norm": 0.320420503616333, | |
| "learning_rate": 0.00011658291457286432, | |
| "loss": 0.9266, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.18727758007117437, | |
| "grad_norm": 0.3732924461364746, | |
| "learning_rate": 0.00011638190954773872, | |
| "loss": 0.9703, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.18772241992882563, | |
| "grad_norm": 0.3584558069705963, | |
| "learning_rate": 0.00011618090452261308, | |
| "loss": 0.8832, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.18816725978647686, | |
| "grad_norm": 0.3892074525356293, | |
| "learning_rate": 0.00011597989949748745, | |
| "loss": 0.9617, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.18861209964412812, | |
| "grad_norm": 0.3820159435272217, | |
| "learning_rate": 0.00011577889447236181, | |
| "loss": 0.9092, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.18905693950177935, | |
| "grad_norm": 0.33480194211006165, | |
| "learning_rate": 0.00011557788944723618, | |
| "loss": 1.0104, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.1895017793594306, | |
| "grad_norm": 0.41536465287208557, | |
| "learning_rate": 0.00011537688442211057, | |
| "loss": 0.9157, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.18994661921708186, | |
| "grad_norm": 0.3305935263633728, | |
| "learning_rate": 0.00011517587939698493, | |
| "loss": 0.9622, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.1903914590747331, | |
| "grad_norm": 0.4637777805328369, | |
| "learning_rate": 0.0001149748743718593, | |
| "loss": 0.9317, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.19083629893238435, | |
| "grad_norm": 0.38072511553764343, | |
| "learning_rate": 0.00011477386934673366, | |
| "loss": 1.0061, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.19128113879003558, | |
| "grad_norm": 0.35834434628486633, | |
| "learning_rate": 0.00011457286432160806, | |
| "loss": 1.0108, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.19172597864768684, | |
| "grad_norm": 0.4265679717063904, | |
| "learning_rate": 0.00011437185929648242, | |
| "loss": 0.9451, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.19217081850533807, | |
| "grad_norm": 0.38988345861434937, | |
| "learning_rate": 0.00011417085427135679, | |
| "loss": 1.1026, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.19261565836298933, | |
| "grad_norm": 0.35931700468063354, | |
| "learning_rate": 0.00011396984924623115, | |
| "loss": 0.9296, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.19306049822064056, | |
| "grad_norm": 0.3506704568862915, | |
| "learning_rate": 0.00011376884422110554, | |
| "loss": 0.942, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.19350533807829182, | |
| "grad_norm": 0.38077566027641296, | |
| "learning_rate": 0.00011356783919597991, | |
| "loss": 1.0012, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.19395017793594305, | |
| "grad_norm": 0.3146056532859802, | |
| "learning_rate": 0.00011336683417085427, | |
| "loss": 0.9361, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.1943950177935943, | |
| "grad_norm": 0.42160287499427795, | |
| "learning_rate": 0.00011316582914572864, | |
| "loss": 0.9426, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.19483985765124556, | |
| "grad_norm": 0.33514153957366943, | |
| "learning_rate": 0.00011296482412060303, | |
| "loss": 1.0237, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.1952846975088968, | |
| "grad_norm": 0.46194738149642944, | |
| "learning_rate": 0.0001127638190954774, | |
| "loss": 1.095, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.19572953736654805, | |
| "grad_norm": 0.40086501836776733, | |
| "learning_rate": 0.00011256281407035176, | |
| "loss": 0.9239, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.19617437722419928, | |
| "grad_norm": 0.34976860880851746, | |
| "learning_rate": 0.00011236180904522614, | |
| "loss": 0.8445, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.19661921708185054, | |
| "grad_norm": 0.35818928480148315, | |
| "learning_rate": 0.00011216080402010052, | |
| "loss": 0.9626, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.19706405693950177, | |
| "grad_norm": 0.35546717047691345, | |
| "learning_rate": 0.00011195979899497488, | |
| "loss": 0.981, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.19750889679715303, | |
| "grad_norm": 0.32887890934944153, | |
| "learning_rate": 0.00011175879396984925, | |
| "loss": 0.9953, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.19795373665480426, | |
| "grad_norm": 0.3613591492176056, | |
| "learning_rate": 0.00011155778894472361, | |
| "loss": 0.9798, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.19839857651245552, | |
| "grad_norm": 0.3338494896888733, | |
| "learning_rate": 0.00011135678391959799, | |
| "loss": 1.0195, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.19884341637010677, | |
| "grad_norm": 0.3248537480831146, | |
| "learning_rate": 0.00011115577889447237, | |
| "loss": 0.9145, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.199288256227758, | |
| "grad_norm": 0.35757869482040405, | |
| "learning_rate": 0.00011095477386934675, | |
| "loss": 0.898, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.19973309608540926, | |
| "grad_norm": 0.35583075881004333, | |
| "learning_rate": 0.0001107537688442211, | |
| "loss": 1.0303, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.2001779359430605, | |
| "grad_norm": 0.5167235136032104, | |
| "learning_rate": 0.00011055276381909548, | |
| "loss": 0.8168, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.20062277580071175, | |
| "grad_norm": 0.3596641719341278, | |
| "learning_rate": 0.00011035175879396986, | |
| "loss": 0.9184, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.20106761565836298, | |
| "grad_norm": 0.4046080410480499, | |
| "learning_rate": 0.00011015075376884422, | |
| "loss": 1.0039, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.20151245551601424, | |
| "grad_norm": 0.36985480785369873, | |
| "learning_rate": 0.0001099497487437186, | |
| "loss": 0.9409, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.20195729537366547, | |
| "grad_norm": 0.36848029494285583, | |
| "learning_rate": 0.00010974874371859296, | |
| "loss": 1.0075, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.20240213523131673, | |
| "grad_norm": 0.3421315550804138, | |
| "learning_rate": 0.00010954773869346736, | |
| "loss": 1.0761, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.20284697508896798, | |
| "grad_norm": 0.35291051864624023, | |
| "learning_rate": 0.00010934673366834172, | |
| "loss": 0.9096, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.20329181494661921, | |
| "grad_norm": 0.3597058355808258, | |
| "learning_rate": 0.00010914572864321609, | |
| "loss": 0.9133, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.20373665480427047, | |
| "grad_norm": 0.3638782501220703, | |
| "learning_rate": 0.00010894472361809045, | |
| "loss": 0.9936, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.2041814946619217, | |
| "grad_norm": 0.33646896481513977, | |
| "learning_rate": 0.00010874371859296483, | |
| "loss": 0.9592, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.20462633451957296, | |
| "grad_norm": 0.3791368901729584, | |
| "learning_rate": 0.00010854271356783921, | |
| "loss": 0.9708, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.2050711743772242, | |
| "grad_norm": 0.3495243489742279, | |
| "learning_rate": 0.00010834170854271357, | |
| "loss": 0.9908, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.20551601423487545, | |
| "grad_norm": 0.34329208731651306, | |
| "learning_rate": 0.00010814070351758794, | |
| "loss": 0.9722, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.20596085409252668, | |
| "grad_norm": 0.40529054403305054, | |
| "learning_rate": 0.00010793969849246233, | |
| "loss": 0.9758, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.20640569395017794, | |
| "grad_norm": 0.44128406047821045, | |
| "learning_rate": 0.0001077386934673367, | |
| "loss": 0.9661, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.20685053380782917, | |
| "grad_norm": 0.3696465790271759, | |
| "learning_rate": 0.00010753768844221106, | |
| "loss": 1.0031, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.20729537366548043, | |
| "grad_norm": 0.36873361468315125, | |
| "learning_rate": 0.00010733668341708543, | |
| "loss": 0.9141, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.20774021352313168, | |
| "grad_norm": 0.40861397981643677, | |
| "learning_rate": 0.00010713567839195982, | |
| "loss": 0.9727, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.20818505338078291, | |
| "grad_norm": 0.3468811511993408, | |
| "learning_rate": 0.00010693467336683418, | |
| "loss": 1.0182, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.20862989323843417, | |
| "grad_norm": 0.37425243854522705, | |
| "learning_rate": 0.00010673366834170855, | |
| "loss": 0.9268, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.2090747330960854, | |
| "grad_norm": 0.4296034574508667, | |
| "learning_rate": 0.00010653266331658291, | |
| "loss": 1.116, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.20951957295373666, | |
| "grad_norm": 0.3852652609348297, | |
| "learning_rate": 0.00010633165829145728, | |
| "loss": 0.9378, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.2099644128113879, | |
| "grad_norm": 0.34109607338905334, | |
| "learning_rate": 0.00010613065326633167, | |
| "loss": 1.004, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.21040925266903915, | |
| "grad_norm": 0.4025616943836212, | |
| "learning_rate": 0.00010592964824120604, | |
| "loss": 0.96, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.21085409252669038, | |
| "grad_norm": 0.3672776222229004, | |
| "learning_rate": 0.0001057286432160804, | |
| "loss": 0.8744, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.21129893238434164, | |
| "grad_norm": 0.36405524611473083, | |
| "learning_rate": 0.00010552763819095478, | |
| "loss": 0.8957, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.2117437722419929, | |
| "grad_norm": 0.41747644543647766, | |
| "learning_rate": 0.00010532663316582916, | |
| "loss": 1.0165, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.21218861209964412, | |
| "grad_norm": 0.3478085994720459, | |
| "learning_rate": 0.00010512562814070352, | |
| "loss": 0.9776, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.21263345195729538, | |
| "grad_norm": 0.365633487701416, | |
| "learning_rate": 0.0001049246231155779, | |
| "loss": 0.9826, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.2130782918149466, | |
| "grad_norm": 0.3211476802825928, | |
| "learning_rate": 0.00010472361809045225, | |
| "loss": 0.9228, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.21352313167259787, | |
| "grad_norm": 0.3374654948711395, | |
| "learning_rate": 0.00010452261306532664, | |
| "loss": 0.9379, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.2139679715302491, | |
| "grad_norm": 0.34640708565711975, | |
| "learning_rate": 0.00010432160804020101, | |
| "loss": 0.8678, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.21441281138790036, | |
| "grad_norm": 0.38176846504211426, | |
| "learning_rate": 0.00010412060301507539, | |
| "loss": 0.9597, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.2148576512455516, | |
| "grad_norm": 0.34589311480522156, | |
| "learning_rate": 0.00010391959798994975, | |
| "loss": 0.9423, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.21530249110320285, | |
| "grad_norm": 0.4062221348285675, | |
| "learning_rate": 0.00010371859296482413, | |
| "loss": 0.9926, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.2157473309608541, | |
| "grad_norm": 0.3403069078922272, | |
| "learning_rate": 0.0001035175879396985, | |
| "loss": 0.9953, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.21619217081850534, | |
| "grad_norm": 0.3341009318828583, | |
| "learning_rate": 0.00010331658291457286, | |
| "loss": 0.9807, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.2166370106761566, | |
| "grad_norm": 0.29718342423439026, | |
| "learning_rate": 0.00010311557788944724, | |
| "loss": 0.9296, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.21708185053380782, | |
| "grad_norm": 0.4219815135002136, | |
| "learning_rate": 0.00010291457286432162, | |
| "loss": 1.0007, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.21752669039145908, | |
| "grad_norm": 0.30951768159866333, | |
| "learning_rate": 0.00010271356783919598, | |
| "loss": 0.9686, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.2179715302491103, | |
| "grad_norm": 0.38819047808647156, | |
| "learning_rate": 0.00010251256281407036, | |
| "loss": 1.1859, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.21841637010676157, | |
| "grad_norm": 0.3185841739177704, | |
| "learning_rate": 0.00010231155778894473, | |
| "loss": 0.9523, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.2188612099644128, | |
| "grad_norm": 0.3971594274044037, | |
| "learning_rate": 0.00010211055276381909, | |
| "loss": 1.0133, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.21930604982206406, | |
| "grad_norm": 0.31492552161216736, | |
| "learning_rate": 0.00010190954773869348, | |
| "loss": 0.8269, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.2197508896797153, | |
| "grad_norm": 0.3949122130870819, | |
| "learning_rate": 0.00010170854271356785, | |
| "loss": 1.0379, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.22019572953736655, | |
| "grad_norm": 0.30094975233078003, | |
| "learning_rate": 0.00010150753768844221, | |
| "loss": 0.9071, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.2206405693950178, | |
| "grad_norm": 0.3560626208782196, | |
| "learning_rate": 0.00010130653266331658, | |
| "loss": 0.9853, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.22108540925266904, | |
| "grad_norm": 0.4022250771522522, | |
| "learning_rate": 0.00010110552763819097, | |
| "loss": 0.9823, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.2215302491103203, | |
| "grad_norm": 0.3849669098854065, | |
| "learning_rate": 0.00010090452261306533, | |
| "loss": 0.9176, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.22197508896797152, | |
| "grad_norm": 0.47436994314193726, | |
| "learning_rate": 0.0001007035175879397, | |
| "loss": 0.9948, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.22241992882562278, | |
| "grad_norm": 0.32288622856140137, | |
| "learning_rate": 0.00010050251256281407, | |
| "loss": 0.9409, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.222864768683274, | |
| "grad_norm": 0.33923619985580444, | |
| "learning_rate": 0.00010030150753768846, | |
| "loss": 0.9096, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.22330960854092527, | |
| "grad_norm": 0.3114798665046692, | |
| "learning_rate": 0.00010010050251256282, | |
| "loss": 0.9112, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.2237544483985765, | |
| "grad_norm": 0.37581518292427063, | |
| "learning_rate": 9.989949748743719e-05, | |
| "loss": 0.9831, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.22419928825622776, | |
| "grad_norm": 0.33060386776924133, | |
| "learning_rate": 9.969849246231156e-05, | |
| "loss": 0.8942, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.22464412811387902, | |
| "grad_norm": 0.32445380091667175, | |
| "learning_rate": 9.949748743718594e-05, | |
| "loss": 0.9665, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.22508896797153025, | |
| "grad_norm": 0.3652149438858032, | |
| "learning_rate": 9.929648241206031e-05, | |
| "loss": 1.0325, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.2255338078291815, | |
| "grad_norm": 0.3789691925048828, | |
| "learning_rate": 9.909547738693468e-05, | |
| "loss": 0.9488, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.22597864768683273, | |
| "grad_norm": 0.35577788949012756, | |
| "learning_rate": 9.889447236180906e-05, | |
| "loss": 0.9324, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.226423487544484, | |
| "grad_norm": 0.37785062193870544, | |
| "learning_rate": 9.869346733668342e-05, | |
| "loss": 0.9934, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.22686832740213522, | |
| "grad_norm": 0.3577893078327179, | |
| "learning_rate": 9.84924623115578e-05, | |
| "loss": 1.0183, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.22731316725978648, | |
| "grad_norm": 0.3955710232257843, | |
| "learning_rate": 9.829145728643216e-05, | |
| "loss": 0.9138, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.2277580071174377, | |
| "grad_norm": 0.333099365234375, | |
| "learning_rate": 9.809045226130655e-05, | |
| "loss": 0.9533, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.22820284697508897, | |
| "grad_norm": 0.358026921749115, | |
| "learning_rate": 9.788944723618091e-05, | |
| "loss": 0.9919, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.22864768683274023, | |
| "grad_norm": 0.3610975742340088, | |
| "learning_rate": 9.768844221105528e-05, | |
| "loss": 1.0059, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.22909252669039146, | |
| "grad_norm": 0.32842594385147095, | |
| "learning_rate": 9.748743718592965e-05, | |
| "loss": 0.9877, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.22953736654804271, | |
| "grad_norm": 0.4672807455062866, | |
| "learning_rate": 9.728643216080403e-05, | |
| "loss": 1.0148, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.22998220640569395, | |
| "grad_norm": 0.3504094183444977, | |
| "learning_rate": 9.70854271356784e-05, | |
| "loss": 0.8774, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.2304270462633452, | |
| "grad_norm": 0.3299584984779358, | |
| "learning_rate": 9.688442211055276e-05, | |
| "loss": 0.9018, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.23087188612099643, | |
| "grad_norm": 0.3598071336746216, | |
| "learning_rate": 9.668341708542715e-05, | |
| "loss": 0.9, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.2313167259786477, | |
| "grad_norm": 0.366554856300354, | |
| "learning_rate": 9.64824120603015e-05, | |
| "loss": 0.8612, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.23176156583629892, | |
| "grad_norm": 0.3408771753311157, | |
| "learning_rate": 9.628140703517589e-05, | |
| "loss": 0.9482, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.23220640569395018, | |
| "grad_norm": 0.4127398133277893, | |
| "learning_rate": 9.608040201005025e-05, | |
| "loss": 0.9427, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.2326512455516014, | |
| "grad_norm": 0.4046645164489746, | |
| "learning_rate": 9.587939698492462e-05, | |
| "loss": 1.0469, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.23309608540925267, | |
| "grad_norm": 0.33988115191459656, | |
| "learning_rate": 9.5678391959799e-05, | |
| "loss": 0.888, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.23354092526690393, | |
| "grad_norm": 0.35615596175193787, | |
| "learning_rate": 9.547738693467337e-05, | |
| "loss": 0.9522, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.23398576512455516, | |
| "grad_norm": 0.41455528140068054, | |
| "learning_rate": 9.527638190954774e-05, | |
| "loss": 0.9495, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.23443060498220641, | |
| "grad_norm": 0.4435769021511078, | |
| "learning_rate": 9.507537688442212e-05, | |
| "loss": 0.9251, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.23487544483985764, | |
| "grad_norm": 0.34414413571357727, | |
| "learning_rate": 9.487437185929649e-05, | |
| "loss": 0.9746, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.2353202846975089, | |
| "grad_norm": 0.42099055647850037, | |
| "learning_rate": 9.467336683417086e-05, | |
| "loss": 1.0079, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.23576512455516013, | |
| "grad_norm": 0.3634389042854309, | |
| "learning_rate": 9.447236180904523e-05, | |
| "loss": 0.8452, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.2362099644128114, | |
| "grad_norm": 0.45584288239479065, | |
| "learning_rate": 9.427135678391961e-05, | |
| "loss": 0.9618, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.23665480427046262, | |
| "grad_norm": 0.3558484613895416, | |
| "learning_rate": 9.407035175879397e-05, | |
| "loss": 1.007, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.23709964412811388, | |
| "grad_norm": 0.41819700598716736, | |
| "learning_rate": 9.386934673366835e-05, | |
| "loss": 0.8806, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.23754448398576514, | |
| "grad_norm": 0.39965230226516724, | |
| "learning_rate": 9.366834170854271e-05, | |
| "loss": 1.0749, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.23798932384341637, | |
| "grad_norm": 0.35675838589668274, | |
| "learning_rate": 9.34673366834171e-05, | |
| "loss": 1.0074, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.23843416370106763, | |
| "grad_norm": 0.37170112133026123, | |
| "learning_rate": 9.326633165829146e-05, | |
| "loss": 0.9623, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.23887900355871886, | |
| "grad_norm": 0.35851290822029114, | |
| "learning_rate": 9.306532663316585e-05, | |
| "loss": 0.9361, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.2393238434163701, | |
| "grad_norm": 0.4395543038845062, | |
| "learning_rate": 9.28643216080402e-05, | |
| "loss": 0.871, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.23976868327402134, | |
| "grad_norm": 0.2868823707103729, | |
| "learning_rate": 9.266331658291458e-05, | |
| "loss": 0.9066, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.2402135231316726, | |
| "grad_norm": 0.3684181869029999, | |
| "learning_rate": 9.246231155778895e-05, | |
| "loss": 1.0657, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.24065836298932383, | |
| "grad_norm": 0.347599059343338, | |
| "learning_rate": 9.226130653266331e-05, | |
| "loss": 1.0081, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.2411032028469751, | |
| "grad_norm": 0.32958024740219116, | |
| "learning_rate": 9.20603015075377e-05, | |
| "loss": 0.9427, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.24154804270462635, | |
| "grad_norm": 0.3672914505004883, | |
| "learning_rate": 9.185929648241206e-05, | |
| "loss": 0.994, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.24199288256227758, | |
| "grad_norm": 0.3725748658180237, | |
| "learning_rate": 9.165829145728644e-05, | |
| "loss": 0.9835, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.24243772241992884, | |
| "grad_norm": 0.39337027072906494, | |
| "learning_rate": 9.14572864321608e-05, | |
| "loss": 0.8137, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.24288256227758007, | |
| "grad_norm": 0.32661673426628113, | |
| "learning_rate": 9.125628140703519e-05, | |
| "loss": 1.0025, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.24332740213523132, | |
| "grad_norm": 0.3912467956542969, | |
| "learning_rate": 9.105527638190955e-05, | |
| "loss": 0.9087, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.24377224199288255, | |
| "grad_norm": 0.37240296602249146, | |
| "learning_rate": 9.085427135678392e-05, | |
| "loss": 0.9275, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.2442170818505338, | |
| "grad_norm": 0.3758089244365692, | |
| "learning_rate": 9.06532663316583e-05, | |
| "loss": 0.8224, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.24466192170818504, | |
| "grad_norm": 0.3426155745983124, | |
| "learning_rate": 9.045226130653267e-05, | |
| "loss": 0.8894, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2451067615658363, | |
| "grad_norm": 0.3958059549331665, | |
| "learning_rate": 9.025125628140704e-05, | |
| "loss": 0.9429, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.24555160142348753, | |
| "grad_norm": 0.4533610939979553, | |
| "learning_rate": 9.005025125628141e-05, | |
| "loss": 1.2454, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.2459964412811388, | |
| "grad_norm": 0.36734533309936523, | |
| "learning_rate": 8.984924623115579e-05, | |
| "loss": 1.0402, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.24644128113879005, | |
| "grad_norm": 0.42336615920066833, | |
| "learning_rate": 8.964824120603016e-05, | |
| "loss": 1.1425, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.24688612099644128, | |
| "grad_norm": 0.44834333658218384, | |
| "learning_rate": 8.944723618090453e-05, | |
| "loss": 0.8959, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.24733096085409254, | |
| "grad_norm": 0.3234824240207672, | |
| "learning_rate": 8.92462311557789e-05, | |
| "loss": 0.9743, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.24777580071174377, | |
| "grad_norm": 0.3193085491657257, | |
| "learning_rate": 8.904522613065326e-05, | |
| "loss": 0.9685, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.24822064056939502, | |
| "grad_norm": 0.34897491335868835, | |
| "learning_rate": 8.884422110552765e-05, | |
| "loss": 0.9132, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.24866548042704625, | |
| "grad_norm": 0.3368080258369446, | |
| "learning_rate": 8.864321608040201e-05, | |
| "loss": 0.9678, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.2491103202846975, | |
| "grad_norm": 0.3948473036289215, | |
| "learning_rate": 8.84422110552764e-05, | |
| "loss": 0.993, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.24955516014234874, | |
| "grad_norm": 0.40010324120521545, | |
| "learning_rate": 8.824120603015076e-05, | |
| "loss": 0.9779, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.40043747425079346, | |
| "learning_rate": 8.804020100502513e-05, | |
| "loss": 0.9633, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.25044483985765126, | |
| "grad_norm": 0.3239743709564209, | |
| "learning_rate": 8.78391959798995e-05, | |
| "loss": 0.939, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.2508896797153025, | |
| "grad_norm": 0.3126469552516937, | |
| "learning_rate": 8.763819095477387e-05, | |
| "loss": 0.8804, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.2513345195729537, | |
| "grad_norm": 0.32982298731803894, | |
| "learning_rate": 8.743718592964825e-05, | |
| "loss": 0.8698, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.251779359430605, | |
| "grad_norm": 0.4134942293167114, | |
| "learning_rate": 8.723618090452261e-05, | |
| "loss": 0.9995, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.25222419928825623, | |
| "grad_norm": 0.34916219115257263, | |
| "learning_rate": 8.7035175879397e-05, | |
| "loss": 0.9463, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.2526690391459075, | |
| "grad_norm": 0.3170456886291504, | |
| "learning_rate": 8.683417085427135e-05, | |
| "loss": 0.9028, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.2531138790035587, | |
| "grad_norm": 0.377886027097702, | |
| "learning_rate": 8.663316582914574e-05, | |
| "loss": 0.9503, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.25355871886120995, | |
| "grad_norm": 0.3546507656574249, | |
| "learning_rate": 8.64321608040201e-05, | |
| "loss": 0.9509, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.2540035587188612, | |
| "grad_norm": 0.38051891326904297, | |
| "learning_rate": 8.623115577889449e-05, | |
| "loss": 0.8949, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.25444839857651247, | |
| "grad_norm": 0.2983281910419464, | |
| "learning_rate": 8.603015075376884e-05, | |
| "loss": 1.0411, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.2548932384341637, | |
| "grad_norm": 0.3371431529521942, | |
| "learning_rate": 8.582914572864322e-05, | |
| "loss": 0.93, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.25533807829181493, | |
| "grad_norm": 0.39484888315200806, | |
| "learning_rate": 8.562814070351759e-05, | |
| "loss": 0.9633, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.2557829181494662, | |
| "grad_norm": 0.34015023708343506, | |
| "learning_rate": 8.542713567839196e-05, | |
| "loss": 0.944, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.25622775800711745, | |
| "grad_norm": 0.32014408707618713, | |
| "learning_rate": 8.522613065326634e-05, | |
| "loss": 0.829, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.2566725978647687, | |
| "grad_norm": 0.42777976393699646, | |
| "learning_rate": 8.502512562814071e-05, | |
| "loss": 1.0107, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.2571174377224199, | |
| "grad_norm": 0.35483554005622864, | |
| "learning_rate": 8.482412060301508e-05, | |
| "loss": 0.9974, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.25756227758007116, | |
| "grad_norm": 0.41365087032318115, | |
| "learning_rate": 8.462311557788946e-05, | |
| "loss": 1.062, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.2580071174377224, | |
| "grad_norm": 0.37708839774131775, | |
| "learning_rate": 8.442211055276383e-05, | |
| "loss": 0.9283, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.2584519572953737, | |
| "grad_norm": 0.40330421924591064, | |
| "learning_rate": 8.42211055276382e-05, | |
| "loss": 0.9482, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.25889679715302494, | |
| "grad_norm": 0.2943516969680786, | |
| "learning_rate": 8.402010050251256e-05, | |
| "loss": 0.9211, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.25934163701067614, | |
| "grad_norm": 0.3328346610069275, | |
| "learning_rate": 8.381909547738695e-05, | |
| "loss": 0.9264, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.2597864768683274, | |
| "grad_norm": 0.4220837354660034, | |
| "learning_rate": 8.36180904522613e-05, | |
| "loss": 0.9598, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.26023131672597866, | |
| "grad_norm": 0.36552560329437256, | |
| "learning_rate": 8.341708542713568e-05, | |
| "loss": 0.9275, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.2606761565836299, | |
| "grad_norm": 0.36409202218055725, | |
| "learning_rate": 8.321608040201005e-05, | |
| "loss": 0.9246, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.2611209964412811, | |
| "grad_norm": 0.3256109952926636, | |
| "learning_rate": 8.301507537688443e-05, | |
| "loss": 0.9725, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.2615658362989324, | |
| "grad_norm": 0.31421205401420593, | |
| "learning_rate": 8.28140703517588e-05, | |
| "loss": 0.9276, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.26201067615658363, | |
| "grad_norm": 0.33592212200164795, | |
| "learning_rate": 8.261306532663317e-05, | |
| "loss": 0.964, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.2624555160142349, | |
| "grad_norm": 0.49559640884399414, | |
| "learning_rate": 8.241206030150754e-05, | |
| "loss": 0.9688, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.2629003558718861, | |
| "grad_norm": 0.37904122471809387, | |
| "learning_rate": 8.22110552763819e-05, | |
| "loss": 0.8688, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.26334519572953735, | |
| "grad_norm": 0.38418570160865784, | |
| "learning_rate": 8.201005025125629e-05, | |
| "loss": 0.966, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.2637900355871886, | |
| "grad_norm": 0.3689751923084259, | |
| "learning_rate": 8.180904522613065e-05, | |
| "loss": 0.8973, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.26423487544483987, | |
| "grad_norm": 0.3859994113445282, | |
| "learning_rate": 8.160804020100504e-05, | |
| "loss": 1.034, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.2646797153024911, | |
| "grad_norm": 0.4068204164505005, | |
| "learning_rate": 8.14070351758794e-05, | |
| "loss": 0.931, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.26512455516014233, | |
| "grad_norm": 0.3974169194698334, | |
| "learning_rate": 8.120603015075378e-05, | |
| "loss": 1.055, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.2655693950177936, | |
| "grad_norm": 0.3818078637123108, | |
| "learning_rate": 8.100502512562814e-05, | |
| "loss": 1.0652, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.26601423487544484, | |
| "grad_norm": 0.3733369708061218, | |
| "learning_rate": 8.080402010050251e-05, | |
| "loss": 0.9981, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.2664590747330961, | |
| "grad_norm": 0.3980778455734253, | |
| "learning_rate": 8.060301507537689e-05, | |
| "loss": 0.982, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.2669039145907473, | |
| "grad_norm": 0.3596290051937103, | |
| "learning_rate": 8.040201005025126e-05, | |
| "loss": 0.896, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.26734875444839856, | |
| "grad_norm": 0.330400675535202, | |
| "learning_rate": 8.020100502512563e-05, | |
| "loss": 0.9776, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.2677935943060498, | |
| "grad_norm": 0.3300771415233612, | |
| "learning_rate": 8e-05, | |
| "loss": 1.0576, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.2682384341637011, | |
| "grad_norm": 0.36290502548217773, | |
| "learning_rate": 7.979899497487438e-05, | |
| "loss": 0.9378, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.26868327402135234, | |
| "grad_norm": 0.3487003743648529, | |
| "learning_rate": 7.959798994974875e-05, | |
| "loss": 0.9146, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.26912811387900354, | |
| "grad_norm": 0.3635631501674652, | |
| "learning_rate": 7.939698492462313e-05, | |
| "loss": 0.9758, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.2695729537366548, | |
| "grad_norm": 0.35547807812690735, | |
| "learning_rate": 7.91959798994975e-05, | |
| "loss": 1.0026, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.27001779359430605, | |
| "grad_norm": 0.33966726064682007, | |
| "learning_rate": 7.899497487437186e-05, | |
| "loss": 0.9059, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.2704626334519573, | |
| "grad_norm": 0.3759270906448364, | |
| "learning_rate": 7.879396984924623e-05, | |
| "loss": 1.0088, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.2709074733096085, | |
| "grad_norm": 0.36231812834739685, | |
| "learning_rate": 7.85929648241206e-05, | |
| "loss": 0.9903, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.2713523131672598, | |
| "grad_norm": 0.389737069606781, | |
| "learning_rate": 7.839195979899498e-05, | |
| "loss": 0.8706, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.27179715302491103, | |
| "grad_norm": 0.43408992886543274, | |
| "learning_rate": 7.819095477386935e-05, | |
| "loss": 0.9437, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.2722419928825623, | |
| "grad_norm": 0.29707393050193787, | |
| "learning_rate": 7.798994974874372e-05, | |
| "loss": 0.8361, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.27268683274021355, | |
| "grad_norm": 0.3469286561012268, | |
| "learning_rate": 7.77889447236181e-05, | |
| "loss": 1.0085, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.27313167259786475, | |
| "grad_norm": 0.4895261824131012, | |
| "learning_rate": 7.758793969849247e-05, | |
| "loss": 0.8068, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.273576512455516, | |
| "grad_norm": 0.3655518591403961, | |
| "learning_rate": 7.738693467336684e-05, | |
| "loss": 0.9248, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.27402135231316727, | |
| "grad_norm": 0.36705151200294495, | |
| "learning_rate": 7.71859296482412e-05, | |
| "loss": 1.0244, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.2744661921708185, | |
| "grad_norm": 0.3587312400341034, | |
| "learning_rate": 7.698492462311559e-05, | |
| "loss": 1.0522, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.2749110320284697, | |
| "grad_norm": 0.3758445680141449, | |
| "learning_rate": 7.678391959798995e-05, | |
| "loss": 0.9904, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.275355871886121, | |
| "grad_norm": 0.31862393021583557, | |
| "learning_rate": 7.658291457286433e-05, | |
| "loss": 0.9216, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.27580071174377224, | |
| "grad_norm": 0.4419485926628113, | |
| "learning_rate": 7.638190954773869e-05, | |
| "loss": 0.9049, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.2762455516014235, | |
| "grad_norm": 0.38519737124443054, | |
| "learning_rate": 7.618090452261307e-05, | |
| "loss": 0.9116, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.27669039145907476, | |
| "grad_norm": 0.37228310108184814, | |
| "learning_rate": 7.597989949748744e-05, | |
| "loss": 0.9491, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.27713523131672596, | |
| "grad_norm": 0.3866081237792969, | |
| "learning_rate": 7.577889447236181e-05, | |
| "loss": 0.9477, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.2775800711743772, | |
| "grad_norm": 0.32213568687438965, | |
| "learning_rate": 7.557788944723618e-05, | |
| "loss": 0.9049, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.2780249110320285, | |
| "grad_norm": 0.34658104181289673, | |
| "learning_rate": 7.537688442211056e-05, | |
| "loss": 0.9486, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.27846975088967973, | |
| "grad_norm": 0.42399680614471436, | |
| "learning_rate": 7.517587939698493e-05, | |
| "loss": 1.0776, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.27891459074733094, | |
| "grad_norm": 0.3882311284542084, | |
| "learning_rate": 7.49748743718593e-05, | |
| "loss": 0.9711, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.2793594306049822, | |
| "grad_norm": 0.34879353642463684, | |
| "learning_rate": 7.477386934673368e-05, | |
| "loss": 0.9253, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.27980427046263345, | |
| "grad_norm": 0.3509921431541443, | |
| "learning_rate": 7.457286432160805e-05, | |
| "loss": 0.9711, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.2802491103202847, | |
| "grad_norm": 0.32417482137680054, | |
| "learning_rate": 7.437185929648241e-05, | |
| "loss": 0.9612, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.28069395017793597, | |
| "grad_norm": 0.3800636827945709, | |
| "learning_rate": 7.417085427135678e-05, | |
| "loss": 1.0243, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.28113879003558717, | |
| "grad_norm": 0.38342657685279846, | |
| "learning_rate": 7.396984924623115e-05, | |
| "loss": 0.9895, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.28158362989323843, | |
| "grad_norm": 0.40875983238220215, | |
| "learning_rate": 7.376884422110553e-05, | |
| "loss": 0.9428, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.2820284697508897, | |
| "grad_norm": 0.34101223945617676, | |
| "learning_rate": 7.35678391959799e-05, | |
| "loss": 0.977, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.28247330960854095, | |
| "grad_norm": 0.3537449836730957, | |
| "learning_rate": 7.336683417085427e-05, | |
| "loss": 0.9458, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.28291814946619215, | |
| "grad_norm": 0.3816027343273163, | |
| "learning_rate": 7.316582914572865e-05, | |
| "loss": 1.0429, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.2833629893238434, | |
| "grad_norm": 0.39149439334869385, | |
| "learning_rate": 7.296482412060302e-05, | |
| "loss": 0.974, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.28380782918149466, | |
| "grad_norm": 0.4129406809806824, | |
| "learning_rate": 7.276381909547739e-05, | |
| "loss": 0.988, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.2842526690391459, | |
| "grad_norm": 0.36804699897766113, | |
| "learning_rate": 7.256281407035177e-05, | |
| "loss": 0.9375, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.2846975088967972, | |
| "grad_norm": 0.40363621711730957, | |
| "learning_rate": 7.236180904522614e-05, | |
| "loss": 0.9214, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2851423487544484, | |
| "grad_norm": 0.3833264708518982, | |
| "learning_rate": 7.21608040201005e-05, | |
| "loss": 0.9454, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.28558718861209964, | |
| "grad_norm": 0.38440853357315063, | |
| "learning_rate": 7.195979899497488e-05, | |
| "loss": 0.8841, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.2860320284697509, | |
| "grad_norm": 0.42536038160324097, | |
| "learning_rate": 7.175879396984924e-05, | |
| "loss": 0.9665, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.28647686832740216, | |
| "grad_norm": 0.33145928382873535, | |
| "learning_rate": 7.155778894472363e-05, | |
| "loss": 0.9118, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.28692170818505336, | |
| "grad_norm": 0.37018072605133057, | |
| "learning_rate": 7.135678391959799e-05, | |
| "loss": 0.9424, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.2873665480427046, | |
| "grad_norm": 0.37953630089759827, | |
| "learning_rate": 7.115577889447236e-05, | |
| "loss": 0.9889, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.2878113879003559, | |
| "grad_norm": 0.3908143937587738, | |
| "learning_rate": 7.095477386934674e-05, | |
| "loss": 0.9815, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.28825622775800713, | |
| "grad_norm": 0.38611525297164917, | |
| "learning_rate": 7.075376884422111e-05, | |
| "loss": 1.0179, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.28870106761565834, | |
| "grad_norm": 0.4868585169315338, | |
| "learning_rate": 7.055276381909548e-05, | |
| "loss": 1.0551, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.2891459074733096, | |
| "grad_norm": 0.4477649927139282, | |
| "learning_rate": 7.035175879396985e-05, | |
| "loss": 0.8795, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.28959074733096085, | |
| "grad_norm": 0.3561191260814667, | |
| "learning_rate": 7.015075376884423e-05, | |
| "loss": 0.8927, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.2900355871886121, | |
| "grad_norm": 0.4050782024860382, | |
| "learning_rate": 6.99497487437186e-05, | |
| "loss": 0.9032, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.29048042704626337, | |
| "grad_norm": 0.3366767466068268, | |
| "learning_rate": 6.974874371859297e-05, | |
| "loss": 0.9763, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.29092526690391457, | |
| "grad_norm": 0.33202803134918213, | |
| "learning_rate": 6.954773869346733e-05, | |
| "loss": 0.8999, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.29137010676156583, | |
| "grad_norm": 0.38702261447906494, | |
| "learning_rate": 6.93467336683417e-05, | |
| "loss": 0.8882, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.2918149466192171, | |
| "grad_norm": 0.39489811658859253, | |
| "learning_rate": 6.914572864321608e-05, | |
| "loss": 0.8615, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.29225978647686834, | |
| "grad_norm": 0.407174289226532, | |
| "learning_rate": 6.894472361809045e-05, | |
| "loss": 0.8805, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.29270462633451955, | |
| "grad_norm": 0.3787648379802704, | |
| "learning_rate": 6.874371859296482e-05, | |
| "loss": 0.9016, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.2931494661921708, | |
| "grad_norm": 0.3571087121963501, | |
| "learning_rate": 6.85427135678392e-05, | |
| "loss": 0.9484, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.29359430604982206, | |
| "grad_norm": 0.41471996903419495, | |
| "learning_rate": 6.834170854271357e-05, | |
| "loss": 1.0558, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2940391459074733, | |
| "grad_norm": 0.4297381639480591, | |
| "learning_rate": 6.814070351758794e-05, | |
| "loss": 0.9687, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.2944839857651246, | |
| "grad_norm": 0.33687537908554077, | |
| "learning_rate": 6.793969849246232e-05, | |
| "loss": 0.9266, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.2949288256227758, | |
| "grad_norm": 0.34857362508773804, | |
| "learning_rate": 6.773869346733669e-05, | |
| "loss": 0.8977, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.29537366548042704, | |
| "grad_norm": 0.4323669672012329, | |
| "learning_rate": 6.753768844221105e-05, | |
| "loss": 0.9535, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.2958185053380783, | |
| "grad_norm": 0.36021220684051514, | |
| "learning_rate": 6.733668341708544e-05, | |
| "loss": 0.9457, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.29626334519572955, | |
| "grad_norm": 0.3574046492576599, | |
| "learning_rate": 6.71356783919598e-05, | |
| "loss": 0.9377, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.29670818505338076, | |
| "grad_norm": 0.3637276291847229, | |
| "learning_rate": 6.693467336683418e-05, | |
| "loss": 0.9769, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.297153024911032, | |
| "grad_norm": 0.4049037992954254, | |
| "learning_rate": 6.673366834170854e-05, | |
| "loss": 1.0499, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.2975978647686833, | |
| "grad_norm": 0.4278649687767029, | |
| "learning_rate": 6.653266331658293e-05, | |
| "loss": 0.8835, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.29804270462633453, | |
| "grad_norm": 0.37466055154800415, | |
| "learning_rate": 6.633165829145729e-05, | |
| "loss": 0.9493, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.2984875444839858, | |
| "grad_norm": 0.3321222960948944, | |
| "learning_rate": 6.613065326633166e-05, | |
| "loss": 0.9484, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.298932384341637, | |
| "grad_norm": 0.4597262740135193, | |
| "learning_rate": 6.592964824120603e-05, | |
| "loss": 1.0418, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.29937722419928825, | |
| "grad_norm": 0.3227452337741852, | |
| "learning_rate": 6.57286432160804e-05, | |
| "loss": 0.9323, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.2998220640569395, | |
| "grad_norm": 0.3192083239555359, | |
| "learning_rate": 6.552763819095478e-05, | |
| "loss": 0.9021, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.30026690391459077, | |
| "grad_norm": 0.4145206809043884, | |
| "learning_rate": 6.532663316582915e-05, | |
| "loss": 0.9431, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.30071174377224197, | |
| "grad_norm": 0.4133388102054596, | |
| "learning_rate": 6.512562814070352e-05, | |
| "loss": 0.9673, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.3011565836298932, | |
| "grad_norm": 0.4261665940284729, | |
| "learning_rate": 6.492462311557788e-05, | |
| "loss": 1.0128, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.3016014234875445, | |
| "grad_norm": 0.33553653955459595, | |
| "learning_rate": 6.472361809045227e-05, | |
| "loss": 0.9103, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.30204626334519574, | |
| "grad_norm": 0.378233939409256, | |
| "learning_rate": 6.452261306532663e-05, | |
| "loss": 1.1017, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.302491103202847, | |
| "grad_norm": 0.39459675550460815, | |
| "learning_rate": 6.4321608040201e-05, | |
| "loss": 0.9056, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.3029359430604982, | |
| "grad_norm": 0.31103044748306274, | |
| "learning_rate": 6.412060301507538e-05, | |
| "loss": 0.896, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.30338078291814946, | |
| "grad_norm": 0.32393592596054077, | |
| "learning_rate": 6.391959798994975e-05, | |
| "loss": 0.9233, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.3038256227758007, | |
| "grad_norm": 0.37299537658691406, | |
| "learning_rate": 6.371859296482412e-05, | |
| "loss": 0.9587, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.304270462633452, | |
| "grad_norm": 0.3141055703163147, | |
| "learning_rate": 6.35175879396985e-05, | |
| "loss": 0.9314, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.3047153024911032, | |
| "grad_norm": 0.32686251401901245, | |
| "learning_rate": 6.331658291457287e-05, | |
| "loss": 0.9325, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.30516014234875444, | |
| "grad_norm": 0.35383760929107666, | |
| "learning_rate": 6.311557788944724e-05, | |
| "loss": 0.9834, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.3056049822064057, | |
| "grad_norm": 0.30796653032302856, | |
| "learning_rate": 6.291457286432161e-05, | |
| "loss": 0.8842, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.30604982206405695, | |
| "grad_norm": 0.31963440775871277, | |
| "learning_rate": 6.271356783919599e-05, | |
| "loss": 0.9372, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.3064946619217082, | |
| "grad_norm": 0.43244582414627075, | |
| "learning_rate": 6.251256281407035e-05, | |
| "loss": 0.939, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.3069395017793594, | |
| "grad_norm": 0.32725387811660767, | |
| "learning_rate": 6.231155778894473e-05, | |
| "loss": 0.8819, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.30738434163701067, | |
| "grad_norm": 0.3326573371887207, | |
| "learning_rate": 6.211055276381909e-05, | |
| "loss": 1.0157, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.30782918149466193, | |
| "grad_norm": 0.3267054855823517, | |
| "learning_rate": 6.190954773869348e-05, | |
| "loss": 0.8806, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.3082740213523132, | |
| "grad_norm": 0.3328150808811188, | |
| "learning_rate": 6.170854271356784e-05, | |
| "loss": 0.9752, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.3087188612099644, | |
| "grad_norm": 0.42537713050842285, | |
| "learning_rate": 6.150753768844222e-05, | |
| "loss": 0.8548, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.30916370106761565, | |
| "grad_norm": 0.339568167924881, | |
| "learning_rate": 6.130653266331658e-05, | |
| "loss": 0.9964, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.3096085409252669, | |
| "grad_norm": 0.40449681878089905, | |
| "learning_rate": 6.110552763819096e-05, | |
| "loss": 0.8698, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.31005338078291816, | |
| "grad_norm": 0.42647475004196167, | |
| "learning_rate": 6.090452261306533e-05, | |
| "loss": 0.9248, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.3104982206405694, | |
| "grad_norm": 0.3729760944843292, | |
| "learning_rate": 6.070351758793971e-05, | |
| "loss": 0.9945, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.3109430604982206, | |
| "grad_norm": 0.34256407618522644, | |
| "learning_rate": 6.0502512562814076e-05, | |
| "loss": 0.8589, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.3113879003558719, | |
| "grad_norm": 0.37434864044189453, | |
| "learning_rate": 6.030150753768844e-05, | |
| "loss": 0.8921, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.31183274021352314, | |
| "grad_norm": 0.3740074336528778, | |
| "learning_rate": 6.0100502512562815e-05, | |
| "loss": 0.9538, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.3122775800711744, | |
| "grad_norm": 0.2955409288406372, | |
| "learning_rate": 5.989949748743718e-05, | |
| "loss": 0.8796, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.3127224199288256, | |
| "grad_norm": 0.41662129759788513, | |
| "learning_rate": 5.969849246231156e-05, | |
| "loss": 0.976, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.31316725978647686, | |
| "grad_norm": 0.3664419651031494, | |
| "learning_rate": 5.949748743718593e-05, | |
| "loss": 0.9813, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.3136120996441281, | |
| "grad_norm": 0.4082026779651642, | |
| "learning_rate": 5.929648241206031e-05, | |
| "loss": 0.9604, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.3140569395017794, | |
| "grad_norm": 0.34381023049354553, | |
| "learning_rate": 5.909547738693467e-05, | |
| "loss": 0.9438, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.3145017793594306, | |
| "grad_norm": 0.32259446382522583, | |
| "learning_rate": 5.889447236180905e-05, | |
| "loss": 0.8939, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.31494661921708184, | |
| "grad_norm": 0.4534759819507599, | |
| "learning_rate": 5.869346733668342e-05, | |
| "loss": 1.0139, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.3153914590747331, | |
| "grad_norm": 0.3195643723011017, | |
| "learning_rate": 5.849246231155779e-05, | |
| "loss": 0.8886, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.31583629893238435, | |
| "grad_norm": 0.34111788868904114, | |
| "learning_rate": 5.829145728643216e-05, | |
| "loss": 0.9006, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.3162811387900356, | |
| "grad_norm": 0.32145076990127563, | |
| "learning_rate": 5.809045226130654e-05, | |
| "loss": 0.9042, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.3167259786476868, | |
| "grad_norm": 0.38871321082115173, | |
| "learning_rate": 5.7889447236180904e-05, | |
| "loss": 0.8285, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.31717081850533807, | |
| "grad_norm": 0.34648001194000244, | |
| "learning_rate": 5.7688442211055284e-05, | |
| "loss": 0.8948, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.31761565836298933, | |
| "grad_norm": 0.42825043201446533, | |
| "learning_rate": 5.748743718592965e-05, | |
| "loss": 0.9665, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.3180604982206406, | |
| "grad_norm": 0.4382406771183014, | |
| "learning_rate": 5.728643216080403e-05, | |
| "loss": 1.0006, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.3185053380782918, | |
| "grad_norm": 0.3384600579738617, | |
| "learning_rate": 5.7085427135678396e-05, | |
| "loss": 0.893, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.31895017793594305, | |
| "grad_norm": 0.329862505197525, | |
| "learning_rate": 5.688442211055277e-05, | |
| "loss": 0.8064, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.3193950177935943, | |
| "grad_norm": 0.36745166778564453, | |
| "learning_rate": 5.6683417085427135e-05, | |
| "loss": 0.9621, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.31983985765124556, | |
| "grad_norm": 0.36240148544311523, | |
| "learning_rate": 5.6482412060301515e-05, | |
| "loss": 0.8902, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.3202846975088968, | |
| "grad_norm": 0.3071519136428833, | |
| "learning_rate": 5.628140703517588e-05, | |
| "loss": 0.9572, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.320729537366548, | |
| "grad_norm": 0.3393983840942383, | |
| "learning_rate": 5.608040201005026e-05, | |
| "loss": 0.9021, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.3211743772241993, | |
| "grad_norm": 0.3375062942504883, | |
| "learning_rate": 5.587939698492463e-05, | |
| "loss": 0.8505, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.32161921708185054, | |
| "grad_norm": 0.36166590452194214, | |
| "learning_rate": 5.567839195979899e-05, | |
| "loss": 0.9217, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.3220640569395018, | |
| "grad_norm": 0.423556923866272, | |
| "learning_rate": 5.547738693467337e-05, | |
| "loss": 0.8722, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.322508896797153, | |
| "grad_norm": 0.3881194293498993, | |
| "learning_rate": 5.527638190954774e-05, | |
| "loss": 0.9246, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.32295373665480426, | |
| "grad_norm": 0.41174158453941345, | |
| "learning_rate": 5.507537688442211e-05, | |
| "loss": 0.9962, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.3233985765124555, | |
| "grad_norm": 0.3954409062862396, | |
| "learning_rate": 5.487437185929648e-05, | |
| "loss": 0.8989, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.3238434163701068, | |
| "grad_norm": 0.371358722448349, | |
| "learning_rate": 5.467336683417086e-05, | |
| "loss": 0.9655, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.32428825622775803, | |
| "grad_norm": 0.4016554355621338, | |
| "learning_rate": 5.4472361809045224e-05, | |
| "loss": 0.9298, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.32473309608540923, | |
| "grad_norm": 0.45096760988235474, | |
| "learning_rate": 5.4271356783919604e-05, | |
| "loss": 1.0633, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.3251779359430605, | |
| "grad_norm": 0.42669543623924255, | |
| "learning_rate": 5.407035175879397e-05, | |
| "loss": 0.9897, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.32562277580071175, | |
| "grad_norm": 0.3072949945926666, | |
| "learning_rate": 5.386934673366835e-05, | |
| "loss": 0.8538, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.326067615658363, | |
| "grad_norm": 0.3681629002094269, | |
| "learning_rate": 5.3668341708542716e-05, | |
| "loss": 0.9532, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.3265124555160142, | |
| "grad_norm": 0.3667491674423218, | |
| "learning_rate": 5.346733668341709e-05, | |
| "loss": 0.9971, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.32695729537366547, | |
| "grad_norm": 0.3561696410179138, | |
| "learning_rate": 5.3266331658291455e-05, | |
| "loss": 0.9332, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.3274021352313167, | |
| "grad_norm": 0.51035076379776, | |
| "learning_rate": 5.3065326633165835e-05, | |
| "loss": 0.9938, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.327846975088968, | |
| "grad_norm": 0.30791398882865906, | |
| "learning_rate": 5.28643216080402e-05, | |
| "loss": 0.9195, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.32829181494661924, | |
| "grad_norm": 0.37262293696403503, | |
| "learning_rate": 5.266331658291458e-05, | |
| "loss": 0.8857, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.32873665480427045, | |
| "grad_norm": 0.3665037751197815, | |
| "learning_rate": 5.246231155778895e-05, | |
| "loss": 0.8634, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.3291814946619217, | |
| "grad_norm": 0.3333563804626465, | |
| "learning_rate": 5.226130653266332e-05, | |
| "loss": 0.9507, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.32962633451957296, | |
| "grad_norm": 0.3755170404911041, | |
| "learning_rate": 5.206030150753769e-05, | |
| "loss": 0.9742, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.3300711743772242, | |
| "grad_norm": 0.3400593101978302, | |
| "learning_rate": 5.1859296482412066e-05, | |
| "loss": 0.8169, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.3305160142348754, | |
| "grad_norm": 0.5976528525352478, | |
| "learning_rate": 5.165829145728643e-05, | |
| "loss": 1.0596, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.3309608540925267, | |
| "grad_norm": 0.3495856523513794, | |
| "learning_rate": 5.145728643216081e-05, | |
| "loss": 1.1273, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.33140569395017794, | |
| "grad_norm": 0.3898337483406067, | |
| "learning_rate": 5.125628140703518e-05, | |
| "loss": 1.0512, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.3318505338078292, | |
| "grad_norm": 0.4057525396347046, | |
| "learning_rate": 5.1055276381909544e-05, | |
| "loss": 1.0474, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.33229537366548045, | |
| "grad_norm": 0.365829199552536, | |
| "learning_rate": 5.0854271356783924e-05, | |
| "loss": 1.0095, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.33274021352313166, | |
| "grad_norm": 0.3495554029941559, | |
| "learning_rate": 5.065326633165829e-05, | |
| "loss": 0.8723, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.3331850533807829, | |
| "grad_norm": 0.3964870572090149, | |
| "learning_rate": 5.045226130653266e-05, | |
| "loss": 0.9775, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.33362989323843417, | |
| "grad_norm": 0.3591833710670471, | |
| "learning_rate": 5.0251256281407036e-05, | |
| "loss": 0.8438, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.33407473309608543, | |
| "grad_norm": 0.4429285526275635, | |
| "learning_rate": 5.005025125628141e-05, | |
| "loss": 0.9144, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.33451957295373663, | |
| "grad_norm": 0.3427349328994751, | |
| "learning_rate": 4.984924623115578e-05, | |
| "loss": 0.8728, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.3349644128113879, | |
| "grad_norm": 0.4042606055736542, | |
| "learning_rate": 4.9648241206030155e-05, | |
| "loss": 1.0932, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.33540925266903915, | |
| "grad_norm": 0.38601040840148926, | |
| "learning_rate": 4.944723618090453e-05, | |
| "loss": 0.8947, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.3358540925266904, | |
| "grad_norm": 0.31526607275009155, | |
| "learning_rate": 4.92462311557789e-05, | |
| "loss": 0.8783, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.33629893238434166, | |
| "grad_norm": 0.3799164593219757, | |
| "learning_rate": 4.9045226130653274e-05, | |
| "loss": 0.894, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.33674377224199287, | |
| "grad_norm": 0.36352670192718506, | |
| "learning_rate": 4.884422110552764e-05, | |
| "loss": 1.0077, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.3371886120996441, | |
| "grad_norm": 0.37971585988998413, | |
| "learning_rate": 4.864321608040201e-05, | |
| "loss": 0.8504, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.3376334519572954, | |
| "grad_norm": 0.3846040666103363, | |
| "learning_rate": 4.844221105527638e-05, | |
| "loss": 0.9923, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.33807829181494664, | |
| "grad_norm": 0.3028171956539154, | |
| "learning_rate": 4.824120603015075e-05, | |
| "loss": 0.9328, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.33852313167259784, | |
| "grad_norm": 0.35201942920684814, | |
| "learning_rate": 4.8040201005025125e-05, | |
| "loss": 0.8622, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.3389679715302491, | |
| "grad_norm": 0.3402314782142639, | |
| "learning_rate": 4.78391959798995e-05, | |
| "loss": 1.0578, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.33941281138790036, | |
| "grad_norm": 0.43474265933036804, | |
| "learning_rate": 4.763819095477387e-05, | |
| "loss": 1.0589, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.3398576512455516, | |
| "grad_norm": 0.4756290912628174, | |
| "learning_rate": 4.7437185929648244e-05, | |
| "loss": 1.1027, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.3403024911032028, | |
| "grad_norm": 0.3077346682548523, | |
| "learning_rate": 4.723618090452262e-05, | |
| "loss": 0.9635, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.3407473309608541, | |
| "grad_norm": 0.36826518177986145, | |
| "learning_rate": 4.703517587939698e-05, | |
| "loss": 0.9687, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.34119217081850534, | |
| "grad_norm": 0.4034232199192047, | |
| "learning_rate": 4.6834170854271356e-05, | |
| "loss": 0.9209, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.3416370106761566, | |
| "grad_norm": 0.41342729330062866, | |
| "learning_rate": 4.663316582914573e-05, | |
| "loss": 0.9213, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.34208185053380785, | |
| "grad_norm": 0.3749755620956421, | |
| "learning_rate": 4.64321608040201e-05, | |
| "loss": 0.9386, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.34252669039145905, | |
| "grad_norm": 0.3411119878292084, | |
| "learning_rate": 4.6231155778894475e-05, | |
| "loss": 0.8805, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.3429715302491103, | |
| "grad_norm": 0.3668997883796692, | |
| "learning_rate": 4.603015075376885e-05, | |
| "loss": 0.9887, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.34341637010676157, | |
| "grad_norm": 0.45816823840141296, | |
| "learning_rate": 4.582914572864322e-05, | |
| "loss": 0.9031, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.34386120996441283, | |
| "grad_norm": 0.3586982786655426, | |
| "learning_rate": 4.5628140703517594e-05, | |
| "loss": 0.9774, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.34430604982206403, | |
| "grad_norm": 0.42974361777305603, | |
| "learning_rate": 4.542713567839196e-05, | |
| "loss": 1.0421, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.3447508896797153, | |
| "grad_norm": 0.36137211322784424, | |
| "learning_rate": 4.522613065326633e-05, | |
| "loss": 1.0171, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.34519572953736655, | |
| "grad_norm": 0.413507342338562, | |
| "learning_rate": 4.5025125628140706e-05, | |
| "loss": 0.9786, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.3456405693950178, | |
| "grad_norm": 0.42441946268081665, | |
| "learning_rate": 4.482412060301508e-05, | |
| "loss": 1.01, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.34608540925266906, | |
| "grad_norm": 0.35072627663612366, | |
| "learning_rate": 4.462311557788945e-05, | |
| "loss": 0.9709, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.34653024911032027, | |
| "grad_norm": 0.37332162261009216, | |
| "learning_rate": 4.4422110552763825e-05, | |
| "loss": 0.9844, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.3469750889679715, | |
| "grad_norm": 0.388919860124588, | |
| "learning_rate": 4.42211055276382e-05, | |
| "loss": 0.8525, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.3474199288256228, | |
| "grad_norm": 0.3552108407020569, | |
| "learning_rate": 4.4020100502512564e-05, | |
| "loss": 0.9713, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.34786476868327404, | |
| "grad_norm": 0.3655258119106293, | |
| "learning_rate": 4.381909547738694e-05, | |
| "loss": 0.9532, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.34830960854092524, | |
| "grad_norm": 0.38755300641059875, | |
| "learning_rate": 4.3618090452261303e-05, | |
| "loss": 0.9346, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.3487544483985765, | |
| "grad_norm": 0.4547751843929291, | |
| "learning_rate": 4.3417085427135676e-05, | |
| "loss": 0.8872, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.34919928825622776, | |
| "grad_norm": 0.2990303039550781, | |
| "learning_rate": 4.321608040201005e-05, | |
| "loss": 0.8979, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.349644128113879, | |
| "grad_norm": 0.3581852614879608, | |
| "learning_rate": 4.301507537688442e-05, | |
| "loss": 0.9466, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.3500889679715303, | |
| "grad_norm": 0.33662378787994385, | |
| "learning_rate": 4.2814070351758795e-05, | |
| "loss": 0.9274, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.3505338078291815, | |
| "grad_norm": 0.34652361273765564, | |
| "learning_rate": 4.261306532663317e-05, | |
| "loss": 0.9791, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.35097864768683273, | |
| "grad_norm": 0.39401039481163025, | |
| "learning_rate": 4.241206030150754e-05, | |
| "loss": 0.9485, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.351423487544484, | |
| "grad_norm": 0.36272600293159485, | |
| "learning_rate": 4.2211055276381914e-05, | |
| "loss": 0.9341, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.35186832740213525, | |
| "grad_norm": 0.3438786566257477, | |
| "learning_rate": 4.201005025125628e-05, | |
| "loss": 0.8061, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.35231316725978645, | |
| "grad_norm": 0.396484911441803, | |
| "learning_rate": 4.180904522613065e-05, | |
| "loss": 0.9506, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.3527580071174377, | |
| "grad_norm": 0.31982484459877014, | |
| "learning_rate": 4.1608040201005026e-05, | |
| "loss": 0.9372, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.35320284697508897, | |
| "grad_norm": 0.323798805475235, | |
| "learning_rate": 4.14070351758794e-05, | |
| "loss": 0.9342, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.3536476868327402, | |
| "grad_norm": 0.3632016181945801, | |
| "learning_rate": 4.120603015075377e-05, | |
| "loss": 0.9682, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.3540925266903915, | |
| "grad_norm": 0.3542345464229584, | |
| "learning_rate": 4.1005025125628145e-05, | |
| "loss": 0.9129, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.3545373665480427, | |
| "grad_norm": 0.48540955781936646, | |
| "learning_rate": 4.080402010050252e-05, | |
| "loss": 1.0319, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.35498220640569395, | |
| "grad_norm": 0.37020549178123474, | |
| "learning_rate": 4.060301507537689e-05, | |
| "loss": 0.8719, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.3554270462633452, | |
| "grad_norm": 0.3740653097629547, | |
| "learning_rate": 4.040201005025126e-05, | |
| "loss": 0.9464, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.35587188612099646, | |
| "grad_norm": 0.3594771921634674, | |
| "learning_rate": 4.020100502512563e-05, | |
| "loss": 0.9592, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.35631672597864766, | |
| "grad_norm": 0.4490291476249695, | |
| "learning_rate": 4e-05, | |
| "loss": 0.9624, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.3567615658362989, | |
| "grad_norm": 0.3087805509567261, | |
| "learning_rate": 3.9798994974874376e-05, | |
| "loss": 0.9905, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.3572064056939502, | |
| "grad_norm": 0.39388784766197205, | |
| "learning_rate": 3.959798994974875e-05, | |
| "loss": 0.9644, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.35765124555160144, | |
| "grad_norm": 0.411502867937088, | |
| "learning_rate": 3.9396984924623115e-05, | |
| "loss": 0.9547, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.3580960854092527, | |
| "grad_norm": 0.3525221049785614, | |
| "learning_rate": 3.919597989949749e-05, | |
| "loss": 0.8875, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.3585409252669039, | |
| "grad_norm": 0.3365210294723511, | |
| "learning_rate": 3.899497487437186e-05, | |
| "loss": 0.913, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.35898576512455516, | |
| "grad_norm": 0.3405802249908447, | |
| "learning_rate": 3.8793969849246234e-05, | |
| "loss": 0.8679, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.3594306049822064, | |
| "grad_norm": 0.33175504207611084, | |
| "learning_rate": 3.85929648241206e-05, | |
| "loss": 0.929, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.35987544483985767, | |
| "grad_norm": 0.3878396451473236, | |
| "learning_rate": 3.8391959798994973e-05, | |
| "loss": 0.9765, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.3603202846975089, | |
| "grad_norm": 0.3880787491798401, | |
| "learning_rate": 3.8190954773869346e-05, | |
| "loss": 0.9282, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.36076512455516013, | |
| "grad_norm": 0.3783409297466278, | |
| "learning_rate": 3.798994974874372e-05, | |
| "loss": 0.9704, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.3612099644128114, | |
| "grad_norm": 0.37962606549263, | |
| "learning_rate": 3.778894472361809e-05, | |
| "loss": 1.0464, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.36165480427046265, | |
| "grad_norm": 0.4055241644382477, | |
| "learning_rate": 3.7587939698492465e-05, | |
| "loss": 0.9668, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.3620996441281139, | |
| "grad_norm": 0.3240121304988861, | |
| "learning_rate": 3.738693467336684e-05, | |
| "loss": 0.9415, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.3625444839857651, | |
| "grad_norm": 0.29463499784469604, | |
| "learning_rate": 3.7185929648241204e-05, | |
| "loss": 0.9489, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.36298932384341637, | |
| "grad_norm": 0.4620193839073181, | |
| "learning_rate": 3.698492462311558e-05, | |
| "loss": 0.8513, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.3634341637010676, | |
| "grad_norm": 0.3686508536338806, | |
| "learning_rate": 3.678391959798995e-05, | |
| "loss": 1.016, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.3638790035587189, | |
| "grad_norm": 0.366397500038147, | |
| "learning_rate": 3.658291457286432e-05, | |
| "loss": 0.8772, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.3643238434163701, | |
| "grad_norm": 0.3382261097431183, | |
| "learning_rate": 3.6381909547738696e-05, | |
| "loss": 0.9158, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.36476868327402134, | |
| "grad_norm": 0.3870478570461273, | |
| "learning_rate": 3.618090452261307e-05, | |
| "loss": 0.9776, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.3652135231316726, | |
| "grad_norm": 0.40006503462791443, | |
| "learning_rate": 3.597989949748744e-05, | |
| "loss": 0.9127, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.36565836298932386, | |
| "grad_norm": 0.41942328214645386, | |
| "learning_rate": 3.5778894472361815e-05, | |
| "loss": 0.8786, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.36610320284697506, | |
| "grad_norm": 0.34577417373657227, | |
| "learning_rate": 3.557788944723618e-05, | |
| "loss": 0.9029, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.3665480427046263, | |
| "grad_norm": 0.3666742146015167, | |
| "learning_rate": 3.5376884422110554e-05, | |
| "loss": 0.9616, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.3669928825622776, | |
| "grad_norm": 0.41306397318840027, | |
| "learning_rate": 3.517587939698493e-05, | |
| "loss": 0.8994, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.36743772241992884, | |
| "grad_norm": 0.354464054107666, | |
| "learning_rate": 3.49748743718593e-05, | |
| "loss": 0.8398, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.3678825622775801, | |
| "grad_norm": 0.40960076451301575, | |
| "learning_rate": 3.4773869346733667e-05, | |
| "loss": 1.0062, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.3683274021352313, | |
| "grad_norm": 0.35255295038223267, | |
| "learning_rate": 3.457286432160804e-05, | |
| "loss": 1.013, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.36877224199288255, | |
| "grad_norm": 0.320544570684433, | |
| "learning_rate": 3.437185929648241e-05, | |
| "loss": 0.8456, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.3692170818505338, | |
| "grad_norm": 0.3942032754421234, | |
| "learning_rate": 3.4170854271356785e-05, | |
| "loss": 0.9627, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.36966192170818507, | |
| "grad_norm": 0.34646615386009216, | |
| "learning_rate": 3.396984924623116e-05, | |
| "loss": 0.9397, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.3701067615658363, | |
| "grad_norm": 0.3462013006210327, | |
| "learning_rate": 3.3768844221105525e-05, | |
| "loss": 0.9722, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.37055160142348753, | |
| "grad_norm": 0.44638511538505554, | |
| "learning_rate": 3.35678391959799e-05, | |
| "loss": 1.0433, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.3709964412811388, | |
| "grad_norm": 0.3865186870098114, | |
| "learning_rate": 3.336683417085427e-05, | |
| "loss": 0.9243, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.37144128113879005, | |
| "grad_norm": 0.3224859833717346, | |
| "learning_rate": 3.3165829145728643e-05, | |
| "loss": 0.9197, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.3718861209964413, | |
| "grad_norm": 0.3511156439781189, | |
| "learning_rate": 3.2964824120603016e-05, | |
| "loss": 0.951, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.3723309608540925, | |
| "grad_norm": 0.3146522343158722, | |
| "learning_rate": 3.276381909547739e-05, | |
| "loss": 0.9514, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.37277580071174377, | |
| "grad_norm": 0.3094431757926941, | |
| "learning_rate": 3.256281407035176e-05, | |
| "loss": 0.8729, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.373220640569395, | |
| "grad_norm": 0.3724329471588135, | |
| "learning_rate": 3.2361809045226135e-05, | |
| "loss": 0.9897, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.3736654804270463, | |
| "grad_norm": 0.3881984353065491, | |
| "learning_rate": 3.21608040201005e-05, | |
| "loss": 0.9755, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.3741103202846975, | |
| "grad_norm": 0.30279484391212463, | |
| "learning_rate": 3.1959798994974875e-05, | |
| "loss": 0.8903, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.37455516014234874, | |
| "grad_norm": 0.42050573229789734, | |
| "learning_rate": 3.175879396984925e-05, | |
| "loss": 0.9163, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 0.3830379545688629, | |
| "learning_rate": 3.155778894472362e-05, | |
| "loss": 0.9519, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.37544483985765126, | |
| "grad_norm": 0.32881635427474976, | |
| "learning_rate": 3.1356783919597993e-05, | |
| "loss": 0.9225, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.3758896797153025, | |
| "grad_norm": 0.30189692974090576, | |
| "learning_rate": 3.1155778894472366e-05, | |
| "loss": 0.916, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.3763345195729537, | |
| "grad_norm": 0.37163209915161133, | |
| "learning_rate": 3.095477386934674e-05, | |
| "loss": 0.9399, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.376779359430605, | |
| "grad_norm": 0.43418970704078674, | |
| "learning_rate": 3.075376884422111e-05, | |
| "loss": 1.0015, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.37722419928825623, | |
| "grad_norm": 0.3178066313266754, | |
| "learning_rate": 3.055276381909548e-05, | |
| "loss": 0.9131, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.3776690391459075, | |
| "grad_norm": 0.3209547996520996, | |
| "learning_rate": 3.0351758793969855e-05, | |
| "loss": 0.9498, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.3781138790035587, | |
| "grad_norm": 0.4426248371601105, | |
| "learning_rate": 3.015075376884422e-05, | |
| "loss": 0.9335, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.37855871886120995, | |
| "grad_norm": 0.41413775086402893, | |
| "learning_rate": 2.994974874371859e-05, | |
| "loss": 0.8906, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.3790035587188612, | |
| "grad_norm": 0.3665112555027008, | |
| "learning_rate": 2.9748743718592964e-05, | |
| "loss": 0.8849, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.37944839857651247, | |
| "grad_norm": 0.36841294169425964, | |
| "learning_rate": 2.9547738693467337e-05, | |
| "loss": 0.8004, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.3798932384341637, | |
| "grad_norm": 0.4228864014148712, | |
| "learning_rate": 2.934673366834171e-05, | |
| "loss": 1.0018, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.38033807829181493, | |
| "grad_norm": 0.34962666034698486, | |
| "learning_rate": 2.914572864321608e-05, | |
| "loss": 0.9105, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.3807829181494662, | |
| "grad_norm": 0.4448448121547699, | |
| "learning_rate": 2.8944723618090452e-05, | |
| "loss": 0.9391, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.38122775800711745, | |
| "grad_norm": 0.36843836307525635, | |
| "learning_rate": 2.8743718592964825e-05, | |
| "loss": 0.9911, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.3816725978647687, | |
| "grad_norm": 0.35289087891578674, | |
| "learning_rate": 2.8542713567839198e-05, | |
| "loss": 0.9575, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.3821174377224199, | |
| "grad_norm": 0.39545342326164246, | |
| "learning_rate": 2.8341708542713568e-05, | |
| "loss": 0.9074, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.38256227758007116, | |
| "grad_norm": 0.3101719319820404, | |
| "learning_rate": 2.814070351758794e-05, | |
| "loss": 0.8479, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.3830071174377224, | |
| "grad_norm": 0.38031789660453796, | |
| "learning_rate": 2.7939698492462314e-05, | |
| "loss": 1.0281, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.3834519572953737, | |
| "grad_norm": 0.38144952058792114, | |
| "learning_rate": 2.7738693467336686e-05, | |
| "loss": 0.9668, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.38389679715302494, | |
| "grad_norm": 0.41572901606559753, | |
| "learning_rate": 2.7537688442211056e-05, | |
| "loss": 0.9258, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.38434163701067614, | |
| "grad_norm": 0.2982839345932007, | |
| "learning_rate": 2.733668341708543e-05, | |
| "loss": 0.8807, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.3847864768683274, | |
| "grad_norm": 0.3221079111099243, | |
| "learning_rate": 2.7135678391959802e-05, | |
| "loss": 0.9178, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.38523131672597866, | |
| "grad_norm": 0.38755419850349426, | |
| "learning_rate": 2.6934673366834175e-05, | |
| "loss": 0.9183, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.3856761565836299, | |
| "grad_norm": 0.36158278584480286, | |
| "learning_rate": 2.6733668341708545e-05, | |
| "loss": 0.9274, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.3861209964412811, | |
| "grad_norm": 0.308855801820755, | |
| "learning_rate": 2.6532663316582917e-05, | |
| "loss": 0.9158, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.3865658362989324, | |
| "grad_norm": 0.42090025544166565, | |
| "learning_rate": 2.633165829145729e-05, | |
| "loss": 0.9392, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.38701067615658363, | |
| "grad_norm": 0.3310891389846802, | |
| "learning_rate": 2.613065326633166e-05, | |
| "loss": 0.9232, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.3874555160142349, | |
| "grad_norm": 0.3915763199329376, | |
| "learning_rate": 2.5929648241206033e-05, | |
| "loss": 0.8819, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.3879003558718861, | |
| "grad_norm": 0.3351011872291565, | |
| "learning_rate": 2.5728643216080406e-05, | |
| "loss": 0.9757, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.38834519572953735, | |
| "grad_norm": 0.32709652185440063, | |
| "learning_rate": 2.5527638190954772e-05, | |
| "loss": 0.8823, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.3887900355871886, | |
| "grad_norm": 0.34310227632522583, | |
| "learning_rate": 2.5326633165829145e-05, | |
| "loss": 0.8623, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.38923487544483987, | |
| "grad_norm": 0.3940928876399994, | |
| "learning_rate": 2.5125628140703518e-05, | |
| "loss": 0.9409, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.3896797153024911, | |
| "grad_norm": 0.34565237164497375, | |
| "learning_rate": 2.492462311557789e-05, | |
| "loss": 0.9369, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.39012455516014233, | |
| "grad_norm": 0.37526604533195496, | |
| "learning_rate": 2.4723618090452264e-05, | |
| "loss": 0.8391, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.3905693950177936, | |
| "grad_norm": 0.3760671019554138, | |
| "learning_rate": 2.4522613065326637e-05, | |
| "loss": 0.9549, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.39101423487544484, | |
| "grad_norm": 0.35706883668899536, | |
| "learning_rate": 2.4321608040201007e-05, | |
| "loss": 0.7821, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.3914590747330961, | |
| "grad_norm": 0.37825968861579895, | |
| "learning_rate": 2.4120603015075376e-05, | |
| "loss": 1.0344, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.3919039145907473, | |
| "grad_norm": 0.3388988673686981, | |
| "learning_rate": 2.391959798994975e-05, | |
| "loss": 0.9347, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.39234875444839856, | |
| "grad_norm": 0.35239583253860474, | |
| "learning_rate": 2.3718592964824122e-05, | |
| "loss": 0.8896, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.3927935943060498, | |
| "grad_norm": 0.34962180256843567, | |
| "learning_rate": 2.351758793969849e-05, | |
| "loss": 0.9186, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.3932384341637011, | |
| "grad_norm": 0.3848302960395813, | |
| "learning_rate": 2.3316582914572865e-05, | |
| "loss": 1.0796, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.39368327402135234, | |
| "grad_norm": 0.4643072783946991, | |
| "learning_rate": 2.3115577889447238e-05, | |
| "loss": 0.9133, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.39412811387900354, | |
| "grad_norm": 0.3217661678791046, | |
| "learning_rate": 2.291457286432161e-05, | |
| "loss": 0.9077, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.3945729537366548, | |
| "grad_norm": 0.307915061712265, | |
| "learning_rate": 2.271356783919598e-05, | |
| "loss": 0.9465, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.39501779359430605, | |
| "grad_norm": 0.41546520590782166, | |
| "learning_rate": 2.2512562814070353e-05, | |
| "loss": 0.9453, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.3954626334519573, | |
| "grad_norm": 0.44572389125823975, | |
| "learning_rate": 2.2311557788944726e-05, | |
| "loss": 1.1163, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.3959074733096085, | |
| "grad_norm": 0.3532145619392395, | |
| "learning_rate": 2.21105527638191e-05, | |
| "loss": 0.9142, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.3963523131672598, | |
| "grad_norm": 0.3812106251716614, | |
| "learning_rate": 2.190954773869347e-05, | |
| "loss": 0.8557, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.39679715302491103, | |
| "grad_norm": 0.40992099046707153, | |
| "learning_rate": 2.1708542713567838e-05, | |
| "loss": 0.9685, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.3972419928825623, | |
| "grad_norm": 0.35445836186408997, | |
| "learning_rate": 2.150753768844221e-05, | |
| "loss": 0.9416, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.39768683274021355, | |
| "grad_norm": 0.34018823504447937, | |
| "learning_rate": 2.1306532663316584e-05, | |
| "loss": 0.9203, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.39813167259786475, | |
| "grad_norm": 0.37122175097465515, | |
| "learning_rate": 2.1105527638190957e-05, | |
| "loss": 1.1101, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.398576512455516, | |
| "grad_norm": 0.3946356177330017, | |
| "learning_rate": 2.0904522613065327e-05, | |
| "loss": 0.932, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.39902135231316727, | |
| "grad_norm": 0.3920304775238037, | |
| "learning_rate": 2.07035175879397e-05, | |
| "loss": 0.9307, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.3994661921708185, | |
| "grad_norm": 0.35796669125556946, | |
| "learning_rate": 2.0502512562814073e-05, | |
| "loss": 0.9004, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.3999110320284697, | |
| "grad_norm": 0.37431228160858154, | |
| "learning_rate": 2.0301507537688446e-05, | |
| "loss": 0.9462, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.400355871886121, | |
| "grad_norm": 0.43452969193458557, | |
| "learning_rate": 2.0100502512562815e-05, | |
| "loss": 0.9755, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.40080071174377224, | |
| "grad_norm": 0.308403342962265, | |
| "learning_rate": 1.9899497487437188e-05, | |
| "loss": 0.8736, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.4012455516014235, | |
| "grad_norm": 0.425761342048645, | |
| "learning_rate": 1.9698492462311558e-05, | |
| "loss": 1.0335, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.40169039145907476, | |
| "grad_norm": 0.4178042709827423, | |
| "learning_rate": 1.949748743718593e-05, | |
| "loss": 0.8853, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.40213523131672596, | |
| "grad_norm": 0.31192678213119507, | |
| "learning_rate": 1.92964824120603e-05, | |
| "loss": 0.8916, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.4025800711743772, | |
| "grad_norm": 0.4430267810821533, | |
| "learning_rate": 1.9095477386934673e-05, | |
| "loss": 0.962, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.4030249110320285, | |
| "grad_norm": 0.34475943446159363, | |
| "learning_rate": 1.8894472361809046e-05, | |
| "loss": 0.8855, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.40346975088967973, | |
| "grad_norm": 0.36281871795654297, | |
| "learning_rate": 1.869346733668342e-05, | |
| "loss": 0.8925, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.40391459074733094, | |
| "grad_norm": 0.39103028178215027, | |
| "learning_rate": 1.849246231155779e-05, | |
| "loss": 0.86, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.4043594306049822, | |
| "grad_norm": 0.3842105567455292, | |
| "learning_rate": 1.829145728643216e-05, | |
| "loss": 1.0245, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.40480427046263345, | |
| "grad_norm": 0.4096086025238037, | |
| "learning_rate": 1.8090452261306535e-05, | |
| "loss": 0.8985, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.4052491103202847, | |
| "grad_norm": 0.3068806827068329, | |
| "learning_rate": 1.7889447236180908e-05, | |
| "loss": 0.804, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.40569395017793597, | |
| "grad_norm": 0.28686726093292236, | |
| "learning_rate": 1.7688442211055277e-05, | |
| "loss": 1.0117, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.40613879003558717, | |
| "grad_norm": 0.2880958616733551, | |
| "learning_rate": 1.748743718592965e-05, | |
| "loss": 0.8703, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.40658362989323843, | |
| "grad_norm": 0.3917515277862549, | |
| "learning_rate": 1.728643216080402e-05, | |
| "loss": 0.968, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.4070284697508897, | |
| "grad_norm": 0.3814576268196106, | |
| "learning_rate": 1.7085427135678393e-05, | |
| "loss": 0.9305, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.40747330960854095, | |
| "grad_norm": 0.3277512192726135, | |
| "learning_rate": 1.6884422110552762e-05, | |
| "loss": 0.8769, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.40791814946619215, | |
| "grad_norm": 0.3101690709590912, | |
| "learning_rate": 1.6683417085427135e-05, | |
| "loss": 0.8974, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.4083629893238434, | |
| "grad_norm": 0.34977224469184875, | |
| "learning_rate": 1.6482412060301508e-05, | |
| "loss": 0.9463, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.40880782918149466, | |
| "grad_norm": 0.35917431116104126, | |
| "learning_rate": 1.628140703517588e-05, | |
| "loss": 0.9637, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.4092526690391459, | |
| "grad_norm": 0.36873766779899597, | |
| "learning_rate": 1.608040201005025e-05, | |
| "loss": 0.7935, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.4096975088967972, | |
| "grad_norm": 0.348622590303421, | |
| "learning_rate": 1.5879396984924624e-05, | |
| "loss": 0.9165, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.4101423487544484, | |
| "grad_norm": 0.3866550028324127, | |
| "learning_rate": 1.5678391959798997e-05, | |
| "loss": 0.9679, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.41058718861209964, | |
| "grad_norm": 0.3473438620567322, | |
| "learning_rate": 1.547738693467337e-05, | |
| "loss": 0.9228, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.4110320284697509, | |
| "grad_norm": 0.43389132618904114, | |
| "learning_rate": 1.527638190954774e-05, | |
| "loss": 0.9541, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.41147686832740216, | |
| "grad_norm": 0.40540170669555664, | |
| "learning_rate": 1.507537688442211e-05, | |
| "loss": 0.9814, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.41192170818505336, | |
| "grad_norm": 0.3193083107471466, | |
| "learning_rate": 1.4874371859296482e-05, | |
| "loss": 0.9914, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.4123665480427046, | |
| "grad_norm": 0.3414134383201599, | |
| "learning_rate": 1.4673366834170855e-05, | |
| "loss": 1.0247, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.4128113879003559, | |
| "grad_norm": 0.3683622181415558, | |
| "learning_rate": 1.4472361809045226e-05, | |
| "loss": 0.997, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.41325622775800713, | |
| "grad_norm": 0.37435421347618103, | |
| "learning_rate": 1.4271356783919599e-05, | |
| "loss": 0.902, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.41370106761565834, | |
| "grad_norm": 0.31382328271865845, | |
| "learning_rate": 1.407035175879397e-05, | |
| "loss": 0.9035, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.4141459074733096, | |
| "grad_norm": 0.35533666610717773, | |
| "learning_rate": 1.3869346733668343e-05, | |
| "loss": 1.0198, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.41459074733096085, | |
| "grad_norm": 0.3287508189678192, | |
| "learning_rate": 1.3668341708542715e-05, | |
| "loss": 0.8756, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.4150355871886121, | |
| "grad_norm": 0.3902789056301117, | |
| "learning_rate": 1.3467336683417087e-05, | |
| "loss": 0.8743, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.41548042704626337, | |
| "grad_norm": 0.3326834738254547, | |
| "learning_rate": 1.3266331658291459e-05, | |
| "loss": 0.8924, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.41592526690391457, | |
| "grad_norm": 0.3254244923591614, | |
| "learning_rate": 1.306532663316583e-05, | |
| "loss": 0.9863, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.41637010676156583, | |
| "grad_norm": 0.41529661417007446, | |
| "learning_rate": 1.2864321608040203e-05, | |
| "loss": 0.8844, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.4168149466192171, | |
| "grad_norm": 0.31171804666519165, | |
| "learning_rate": 1.2663316582914573e-05, | |
| "loss": 0.9245, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.41725978647686834, | |
| "grad_norm": 0.33847880363464355, | |
| "learning_rate": 1.2462311557788946e-05, | |
| "loss": 0.9414, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.41770462633451955, | |
| "grad_norm": 0.35839855670928955, | |
| "learning_rate": 1.2261306532663318e-05, | |
| "loss": 0.924, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.4181494661921708, | |
| "grad_norm": 0.3934418857097626, | |
| "learning_rate": 1.2060301507537688e-05, | |
| "loss": 0.9718, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.41859430604982206, | |
| "grad_norm": 0.35172393918037415, | |
| "learning_rate": 1.1859296482412061e-05, | |
| "loss": 0.8878, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.4190391459074733, | |
| "grad_norm": 0.4257389307022095, | |
| "learning_rate": 1.1658291457286432e-05, | |
| "loss": 0.9914, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.4194839857651246, | |
| "grad_norm": 0.3889011740684509, | |
| "learning_rate": 1.1457286432160805e-05, | |
| "loss": 0.8339, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.4199288256227758, | |
| "grad_norm": 0.35240477323532104, | |
| "learning_rate": 1.1256281407035177e-05, | |
| "loss": 0.9792, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.42037366548042704, | |
| "grad_norm": 0.3817644715309143, | |
| "learning_rate": 1.105527638190955e-05, | |
| "loss": 0.992, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.4208185053380783, | |
| "grad_norm": 0.36161094903945923, | |
| "learning_rate": 1.0854271356783919e-05, | |
| "loss": 0.8468, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.42126334519572955, | |
| "grad_norm": 0.3254898488521576, | |
| "learning_rate": 1.0653266331658292e-05, | |
| "loss": 0.9998, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.42170818505338076, | |
| "grad_norm": 0.3513511121273041, | |
| "learning_rate": 1.0452261306532663e-05, | |
| "loss": 0.8928, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.422153024911032, | |
| "grad_norm": 0.3499341309070587, | |
| "learning_rate": 1.0251256281407036e-05, | |
| "loss": 0.9137, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.4225978647686833, | |
| "grad_norm": 0.3495025932788849, | |
| "learning_rate": 1.0050251256281408e-05, | |
| "loss": 0.8818, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.42304270462633453, | |
| "grad_norm": 0.3569670021533966, | |
| "learning_rate": 9.849246231155779e-06, | |
| "loss": 0.9912, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.4234875444839858, | |
| "grad_norm": 0.394522100687027, | |
| "learning_rate": 9.64824120603015e-06, | |
| "loss": 0.863, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.423932384341637, | |
| "grad_norm": 0.35868117213249207, | |
| "learning_rate": 9.447236180904523e-06, | |
| "loss": 0.9022, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.42437722419928825, | |
| "grad_norm": 0.33881279826164246, | |
| "learning_rate": 9.246231155778894e-06, | |
| "loss": 0.9634, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.4248220640569395, | |
| "grad_norm": 0.36202844977378845, | |
| "learning_rate": 9.045226130653267e-06, | |
| "loss": 0.9441, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.42526690391459077, | |
| "grad_norm": 0.39875316619873047, | |
| "learning_rate": 8.844221105527639e-06, | |
| "loss": 0.9702, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.42571174377224197, | |
| "grad_norm": 0.3653241693973541, | |
| "learning_rate": 8.64321608040201e-06, | |
| "loss": 0.9436, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.4261565836298932, | |
| "grad_norm": 0.31130754947662354, | |
| "learning_rate": 8.442211055276381e-06, | |
| "loss": 0.8579, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.4266014234875445, | |
| "grad_norm": 0.3955715000629425, | |
| "learning_rate": 8.241206030150754e-06, | |
| "loss": 0.9889, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.42704626334519574, | |
| "grad_norm": 0.35064584016799927, | |
| "learning_rate": 8.040201005025125e-06, | |
| "loss": 0.8877, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.427491103202847, | |
| "grad_norm": 0.38837116956710815, | |
| "learning_rate": 7.839195979899498e-06, | |
| "loss": 0.9255, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.4279359430604982, | |
| "grad_norm": 0.3994256556034088, | |
| "learning_rate": 7.63819095477387e-06, | |
| "loss": 0.9613, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.42838078291814946, | |
| "grad_norm": 0.5106455087661743, | |
| "learning_rate": 7.437185929648241e-06, | |
| "loss": 0.9656, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.4288256227758007, | |
| "grad_norm": 0.46163278818130493, | |
| "learning_rate": 7.236180904522613e-06, | |
| "loss": 1.0239, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.429270462633452, | |
| "grad_norm": 0.37351593375205994, | |
| "learning_rate": 7.035175879396985e-06, | |
| "loss": 0.9637, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.4297153024911032, | |
| "grad_norm": 0.34434255957603455, | |
| "learning_rate": 6.834170854271357e-06, | |
| "loss": 0.9929, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.43016014234875444, | |
| "grad_norm": 0.3309576213359833, | |
| "learning_rate": 6.633165829145729e-06, | |
| "loss": 0.9139, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.4306049822064057, | |
| "grad_norm": 0.5205715298652649, | |
| "learning_rate": 6.4321608040201015e-06, | |
| "loss": 1.0866, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.43104982206405695, | |
| "grad_norm": 0.417214035987854, | |
| "learning_rate": 6.231155778894473e-06, | |
| "loss": 0.9765, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.4314946619217082, | |
| "grad_norm": 0.3578605651855469, | |
| "learning_rate": 6.030150753768844e-06, | |
| "loss": 0.9261, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.4319395017793594, | |
| "grad_norm": 0.35047677159309387, | |
| "learning_rate": 5.829145728643216e-06, | |
| "loss": 0.8318, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.43238434163701067, | |
| "grad_norm": 0.30359870195388794, | |
| "learning_rate": 5.628140703517588e-06, | |
| "loss": 0.9497, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.43282918149466193, | |
| "grad_norm": 0.32898378372192383, | |
| "learning_rate": 5.4271356783919595e-06, | |
| "loss": 0.9873, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.4332740213523132, | |
| "grad_norm": 0.33104610443115234, | |
| "learning_rate": 5.226130653266332e-06, | |
| "loss": 0.9082, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.4337188612099644, | |
| "grad_norm": 0.3245652914047241, | |
| "learning_rate": 5.025125628140704e-06, | |
| "loss": 0.8933, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.43416370106761565, | |
| "grad_norm": 0.3264075517654419, | |
| "learning_rate": 4.824120603015075e-06, | |
| "loss": 0.9778, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.4346085409252669, | |
| "grad_norm": 0.44732666015625, | |
| "learning_rate": 4.623115577889447e-06, | |
| "loss": 0.9509, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.43505338078291816, | |
| "grad_norm": 0.35666903853416443, | |
| "learning_rate": 4.422110552763819e-06, | |
| "loss": 0.9392, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.4354982206405694, | |
| "grad_norm": 0.3573732078075409, | |
| "learning_rate": 4.2211055276381906e-06, | |
| "loss": 0.9323, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.4359430604982206, | |
| "grad_norm": 0.3136303126811981, | |
| "learning_rate": 4.020100502512563e-06, | |
| "loss": 0.959, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.4363879003558719, | |
| "grad_norm": 0.36791837215423584, | |
| "learning_rate": 3.819095477386935e-06, | |
| "loss": 0.9151, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.43683274021352314, | |
| "grad_norm": 0.3033190965652466, | |
| "learning_rate": 3.6180904522613065e-06, | |
| "loss": 0.9223, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.4372775800711744, | |
| "grad_norm": 0.35878488421440125, | |
| "learning_rate": 3.4170854271356786e-06, | |
| "loss": 0.9418, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.4377224199288256, | |
| "grad_norm": 0.3010809123516083, | |
| "learning_rate": 3.2160804020100507e-06, | |
| "loss": 0.8418, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.43816725978647686, | |
| "grad_norm": 0.4400476813316345, | |
| "learning_rate": 3.015075376884422e-06, | |
| "loss": 0.9641, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.4386120996441281, | |
| "grad_norm": 0.34438541531562805, | |
| "learning_rate": 2.814070351758794e-06, | |
| "loss": 0.8904, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.4390569395017794, | |
| "grad_norm": 0.39025887846946716, | |
| "learning_rate": 2.613065326633166e-06, | |
| "loss": 0.9406, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.4395017793594306, | |
| "grad_norm": 0.29612797498703003, | |
| "learning_rate": 2.4120603015075375e-06, | |
| "loss": 0.9153, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.43994661921708184, | |
| "grad_norm": 0.3811575174331665, | |
| "learning_rate": 2.2110552763819096e-06, | |
| "loss": 0.8314, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.4403914590747331, | |
| "grad_norm": 0.3829675614833832, | |
| "learning_rate": 2.0100502512562813e-06, | |
| "loss": 0.8798, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.44083629893238435, | |
| "grad_norm": 0.33976423740386963, | |
| "learning_rate": 1.8090452261306533e-06, | |
| "loss": 0.8793, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.4412811387900356, | |
| "grad_norm": 0.3907322883605957, | |
| "learning_rate": 1.6080402010050254e-06, | |
| "loss": 0.9893, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.4417259786476868, | |
| "grad_norm": 0.33631375432014465, | |
| "learning_rate": 1.407035175879397e-06, | |
| "loss": 0.8798, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.44217081850533807, | |
| "grad_norm": 0.41036009788513184, | |
| "learning_rate": 1.2060301507537688e-06, | |
| "loss": 0.843, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.44261565836298933, | |
| "grad_norm": 0.3805226683616638, | |
| "learning_rate": 1.0050251256281407e-06, | |
| "loss": 1.0452, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.4430604982206406, | |
| "grad_norm": 0.40261736512184143, | |
| "learning_rate": 8.040201005025127e-07, | |
| "loss": 0.8972, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.4435053380782918, | |
| "grad_norm": 0.3762563467025757, | |
| "learning_rate": 6.030150753768844e-07, | |
| "loss": 0.8555, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.44395017793594305, | |
| "grad_norm": 0.3261839747428894, | |
| "learning_rate": 4.0201005025125634e-07, | |
| "loss": 0.8795, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.4443950177935943, | |
| "grad_norm": 0.309962660074234, | |
| "learning_rate": 2.0100502512562817e-07, | |
| "loss": 0.9488, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.44483985765124556, | |
| "grad_norm": 0.3620010018348694, | |
| "learning_rate": 0.0, | |
| "loss": 0.953, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "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": 2.7939011197427712e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |