| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.2668889106775174, | |
| "eval_steps": 341, | |
| "global_step": 1361, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00019609765663300324, | |
| "grad_norm": 18.01366424560547, | |
| "learning_rate": 2e-05, | |
| "loss": 3.0843, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00019609765663300324, | |
| "eval_loss": 1.1017773151397705, | |
| "eval_runtime": 79.0856, | |
| "eval_samples_per_second": 27.16, | |
| "eval_steps_per_second": 13.58, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0003921953132660065, | |
| "grad_norm": 16.80318260192871, | |
| "learning_rate": 4e-05, | |
| "loss": 3.2221, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0005882929698990097, | |
| "grad_norm": 16.65929412841797, | |
| "learning_rate": 6e-05, | |
| "loss": 3.8956, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.000784390626532013, | |
| "grad_norm": 37.782188415527344, | |
| "learning_rate": 8e-05, | |
| "loss": 5.185, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0009804882831650162, | |
| "grad_norm": 19.226940155029297, | |
| "learning_rate": 0.0001, | |
| "loss": 3.1542, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0011765859397980193, | |
| "grad_norm": 26.570402145385742, | |
| "learning_rate": 0.00012, | |
| "loss": 4.5356, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0013726835964310226, | |
| "grad_norm": 22.43348503112793, | |
| "learning_rate": 0.00014, | |
| "loss": 3.8177, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.001568781253064026, | |
| "grad_norm": 29.817941665649414, | |
| "learning_rate": 0.00016, | |
| "loss": 4.4013, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0017648789096970292, | |
| "grad_norm": 18.46044158935547, | |
| "learning_rate": 0.00018, | |
| "loss": 3.374, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0019609765663300325, | |
| "grad_norm": 24.337013244628906, | |
| "learning_rate": 0.0002, | |
| "loss": 3.2509, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0021570742229630358, | |
| "grad_norm": 18.9931640625, | |
| "learning_rate": 0.00019999972962977903, | |
| "loss": 2.933, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0023531718795960386, | |
| "grad_norm": 17.7039852142334, | |
| "learning_rate": 0.00019999891852057812, | |
| "loss": 3.2867, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.002549269536229042, | |
| "grad_norm": 15.128417015075684, | |
| "learning_rate": 0.0001999975666767833, | |
| "loss": 2.2303, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.002745367192862045, | |
| "grad_norm": 8.747583389282227, | |
| "learning_rate": 0.00019999567410570446, | |
| "loss": 1.7348, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0029414648494950485, | |
| "grad_norm": 20.570377349853516, | |
| "learning_rate": 0.00019999324081757555, | |
| "loss": 3.9284, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.003137562506128052, | |
| "grad_norm": 7.86672306060791, | |
| "learning_rate": 0.00019999026682555434, | |
| "loss": 1.8121, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.003333660162761055, | |
| "grad_norm": 10.341567039489746, | |
| "learning_rate": 0.0001999867521457224, | |
| "loss": 1.5864, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0035297578193940584, | |
| "grad_norm": 7.888909339904785, | |
| "learning_rate": 0.00019998269679708504, | |
| "loss": 3.1584, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0037258554760270617, | |
| "grad_norm": 15.346771240234375, | |
| "learning_rate": 0.00019997810080157113, | |
| "loss": 2.0858, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.003921953132660065, | |
| "grad_norm": 11.417168617248535, | |
| "learning_rate": 0.0001999729641840331, | |
| "loss": 3.5646, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.004118050789293068, | |
| "grad_norm": 12.86220932006836, | |
| "learning_rate": 0.00019996728697224675, | |
| "loss": 3.0594, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0043141484459260715, | |
| "grad_norm": 7.418736457824707, | |
| "learning_rate": 0.00019996106919691102, | |
| "loss": 2.1349, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.004510246102559075, | |
| "grad_norm": 4.803528308868408, | |
| "learning_rate": 0.00019995431089164795, | |
| "loss": 2.3151, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.004706343759192077, | |
| "grad_norm": 8.667632102966309, | |
| "learning_rate": 0.00019994701209300245, | |
| "loss": 1.6791, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0049024414158250805, | |
| "grad_norm": 5.3950724601745605, | |
| "learning_rate": 0.00019993917284044202, | |
| "loss": 2.7265, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.005098539072458084, | |
| "grad_norm": 18.16863441467285, | |
| "learning_rate": 0.0001999307931763567, | |
| "loss": 1.7289, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.005294636729091087, | |
| "grad_norm": 10.66887378692627, | |
| "learning_rate": 0.00019992187314605872, | |
| "loss": 2.645, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.00549073438572409, | |
| "grad_norm": 26.42878532409668, | |
| "learning_rate": 0.00019991241279778232, | |
| "loss": 3.4603, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.005686832042357094, | |
| "grad_norm": 10.017987251281738, | |
| "learning_rate": 0.0001999024121826834, | |
| "loss": 1.2001, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.005882929698990097, | |
| "grad_norm": 11.155564308166504, | |
| "learning_rate": 0.00019989187135483933, | |
| "loss": 1.5102, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0060790273556231, | |
| "grad_norm": 3.7590131759643555, | |
| "learning_rate": 0.00019988079037124864, | |
| "loss": 1.0619, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.006275125012256104, | |
| "grad_norm": 8.47985553741455, | |
| "learning_rate": 0.00019986916929183067, | |
| "loss": 2.6256, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.006471222668889107, | |
| "grad_norm": 16.655269622802734, | |
| "learning_rate": 0.00019985700817942533, | |
| "loss": 2.2039, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.00666732032552211, | |
| "grad_norm": 9.003829956054688, | |
| "learning_rate": 0.00019984430709979264, | |
| "loss": 1.7281, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0068634179821551134, | |
| "grad_norm": 36.33080291748047, | |
| "learning_rate": 0.0001998310661216125, | |
| "loss": 3.0575, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.007059515638788117, | |
| "grad_norm": 11.909663200378418, | |
| "learning_rate": 0.00019981728531648423, | |
| "loss": 1.9623, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.00725561329542112, | |
| "grad_norm": 10.967493057250977, | |
| "learning_rate": 0.00019980296475892616, | |
| "loss": 2.8071, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.007451710952054123, | |
| "grad_norm": 11.050918579101562, | |
| "learning_rate": 0.00019978810452637543, | |
| "loss": 1.8019, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.007647808608687127, | |
| "grad_norm": 17.752105712890625, | |
| "learning_rate": 0.00019977270469918727, | |
| "loss": 3.0332, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.00784390626532013, | |
| "grad_norm": 6.67478609085083, | |
| "learning_rate": 0.0001997567653606348, | |
| "loss": 1.32, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.008040003921953132, | |
| "grad_norm": 7.402947902679443, | |
| "learning_rate": 0.00019974028659690843, | |
| "loss": 1.4442, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.008236101578586136, | |
| "grad_norm": 13.006294250488281, | |
| "learning_rate": 0.00019972326849711553, | |
| "loss": 2.2418, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.008432199235219139, | |
| "grad_norm": 5.940983295440674, | |
| "learning_rate": 0.00019970571115327985, | |
| "loss": 0.9049, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.008628296891852143, | |
| "grad_norm": 9.866477012634277, | |
| "learning_rate": 0.00019968761466034103, | |
| "loss": 2.7203, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.008824394548485145, | |
| "grad_norm": 8.239157676696777, | |
| "learning_rate": 0.00019966897911615416, | |
| "loss": 1.9653, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.00902049220511815, | |
| "grad_norm": 16.142181396484375, | |
| "learning_rate": 0.0001996498046214891, | |
| "loss": 2.7509, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.009216589861751152, | |
| "grad_norm": 9.295431137084961, | |
| "learning_rate": 0.00019963009128003018, | |
| "loss": 2.0133, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.009412687518384154, | |
| "grad_norm": 10.362130165100098, | |
| "learning_rate": 0.00019960983919837535, | |
| "loss": 1.716, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.009608785175017159, | |
| "grad_norm": 28.2889461517334, | |
| "learning_rate": 0.00019958904848603584, | |
| "loss": 2.8961, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.009804882831650161, | |
| "grad_norm": 13.199325561523438, | |
| "learning_rate": 0.0001995677192554354, | |
| "loss": 2.62, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.010000980488283165, | |
| "grad_norm": 11.95132827758789, | |
| "learning_rate": 0.00019954585162190985, | |
| "loss": 2.792, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.010197078144916168, | |
| "grad_norm": 16.581575393676758, | |
| "learning_rate": 0.0001995234457037063, | |
| "loss": 2.7239, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.010393175801549172, | |
| "grad_norm": 12.048559188842773, | |
| "learning_rate": 0.00019950050162198258, | |
| "loss": 1.9892, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.010589273458182174, | |
| "grad_norm": 9.297942161560059, | |
| "learning_rate": 0.00019947701950080672, | |
| "loss": 1.8015, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.010785371114815178, | |
| "grad_norm": 6.09962797164917, | |
| "learning_rate": 0.00019945299946715596, | |
| "loss": 1.1493, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.01098146877144818, | |
| "grad_norm": 6.668224811553955, | |
| "learning_rate": 0.00019942844165091633, | |
| "loss": 0.9968, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.011177566428081185, | |
| "grad_norm": 17.000507354736328, | |
| "learning_rate": 0.00019940334618488194, | |
| "loss": 1.5857, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.011373664084714187, | |
| "grad_norm": 11.42551326751709, | |
| "learning_rate": 0.00019937771320475406, | |
| "loss": 1.5236, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.011569761741347192, | |
| "grad_norm": 13.255271911621094, | |
| "learning_rate": 0.00019935154284914065, | |
| "loss": 1.6174, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.011765859397980194, | |
| "grad_norm": 12.664427757263184, | |
| "learning_rate": 0.00019932483525955533, | |
| "loss": 1.476, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.011961957054613198, | |
| "grad_norm": 19.540660858154297, | |
| "learning_rate": 0.00019929759058041687, | |
| "loss": 1.5251, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0121580547112462, | |
| "grad_norm": 10.44942855834961, | |
| "learning_rate": 0.0001992698089590483, | |
| "loss": 1.7865, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.012354152367879203, | |
| "grad_norm": 13.294017791748047, | |
| "learning_rate": 0.00019924149054567606, | |
| "loss": 4.1284, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.012550250024512207, | |
| "grad_norm": 13.700861930847168, | |
| "learning_rate": 0.00019921263549342922, | |
| "loss": 2.8987, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.01274634768114521, | |
| "grad_norm": 51.7520866394043, | |
| "learning_rate": 0.00019918324395833877, | |
| "loss": 1.7335, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.012942445337778214, | |
| "grad_norm": 25.415748596191406, | |
| "learning_rate": 0.00019915331609933657, | |
| "loss": 2.6131, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.013138542994411216, | |
| "grad_norm": 10.575088500976562, | |
| "learning_rate": 0.00019912285207825475, | |
| "loss": 1.202, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.01333464065104422, | |
| "grad_norm": 7.0860772132873535, | |
| "learning_rate": 0.00019909185205982453, | |
| "loss": 1.5077, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.013530738307677223, | |
| "grad_norm": 8.169393539428711, | |
| "learning_rate": 0.00019906031621167553, | |
| "loss": 2.4139, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.013726835964310227, | |
| "grad_norm": 15.750587463378906, | |
| "learning_rate": 0.00019902824470433489, | |
| "loss": 2.8999, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.01392293362094323, | |
| "grad_norm": 19.615734100341797, | |
| "learning_rate": 0.00019899563771122618, | |
| "loss": 3.3154, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.014119031277576233, | |
| "grad_norm": 24.015810012817383, | |
| "learning_rate": 0.0001989624954086686, | |
| "loss": 2.6689, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.014315128934209236, | |
| "grad_norm": 7.955759525299072, | |
| "learning_rate": 0.00019892881797587601, | |
| "loss": 1.8847, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.01451122659084224, | |
| "grad_norm": 13.804747581481934, | |
| "learning_rate": 0.00019889460559495588, | |
| "loss": 2.2221, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.014707324247475242, | |
| "grad_norm": 19.74148941040039, | |
| "learning_rate": 0.0001988598584509084, | |
| "loss": 1.9316, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.014903421904108247, | |
| "grad_norm": 7.9701385498046875, | |
| "learning_rate": 0.00019882457673162543, | |
| "loss": 2.1958, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.015099519560741249, | |
| "grad_norm": 4.7594170570373535, | |
| "learning_rate": 0.00019878876062788954, | |
| "loss": 1.3551, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.015295617217374253, | |
| "grad_norm": 7.796854496002197, | |
| "learning_rate": 0.0001987524103333728, | |
| "loss": 2.616, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.015491714874007256, | |
| "grad_norm": 6.997422218322754, | |
| "learning_rate": 0.00019871552604463602, | |
| "loss": 2.908, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.01568781253064026, | |
| "grad_norm": 6.436347484588623, | |
| "learning_rate": 0.00019867810796112744, | |
| "loss": 3.6826, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.01588391018727326, | |
| "grad_norm": 10.243407249450684, | |
| "learning_rate": 0.00019864015628518175, | |
| "loss": 1.3711, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.016080007843906265, | |
| "grad_norm": 11.830530166625977, | |
| "learning_rate": 0.00019860167122201904, | |
| "loss": 2.2325, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.01627610550053927, | |
| "grad_norm": 7.456768989562988, | |
| "learning_rate": 0.0001985626529797436, | |
| "loss": 1.9991, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.016472203157172273, | |
| "grad_norm": 50.91776657104492, | |
| "learning_rate": 0.00019852310176934288, | |
| "loss": 1.294, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.016668300813805274, | |
| "grad_norm": 6.4661054611206055, | |
| "learning_rate": 0.00019848301780468622, | |
| "loss": 2.4052, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.016864398470438278, | |
| "grad_norm": 5.4749674797058105, | |
| "learning_rate": 0.00019844240130252385, | |
| "loss": 2.1508, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.017060496127071282, | |
| "grad_norm": 9.457857131958008, | |
| "learning_rate": 0.00019840125248248564, | |
| "loss": 2.1732, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.017256593783704286, | |
| "grad_norm": 6.932736396789551, | |
| "learning_rate": 0.00019835957156707988, | |
| "loss": 1.0618, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.017452691440337287, | |
| "grad_norm": 10.447488784790039, | |
| "learning_rate": 0.00019831735878169212, | |
| "loss": 1.1214, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.01764878909697029, | |
| "grad_norm": 7.159199237823486, | |
| "learning_rate": 0.000198274614354584, | |
| "loss": 2.495, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.017844886753603295, | |
| "grad_norm": 8.753946304321289, | |
| "learning_rate": 0.00019823133851689187, | |
| "loss": 2.343, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0180409844102363, | |
| "grad_norm": 9.124870300292969, | |
| "learning_rate": 0.00019818753150262574, | |
| "loss": 1.6556, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0182370820668693, | |
| "grad_norm": 5.470896244049072, | |
| "learning_rate": 0.00019814319354866786, | |
| "loss": 1.2787, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.018433179723502304, | |
| "grad_norm": 8.39749526977539, | |
| "learning_rate": 0.00019809832489477142, | |
| "loss": 1.6804, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.01862927738013531, | |
| "grad_norm": 6.869524955749512, | |
| "learning_rate": 0.0001980529257835594, | |
| "loss": 1.6563, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.01882537503676831, | |
| "grad_norm": 11.144633293151855, | |
| "learning_rate": 0.0001980069964605232, | |
| "loss": 1.9428, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.019021472693401313, | |
| "grad_norm": 11.564960479736328, | |
| "learning_rate": 0.00019796053717402118, | |
| "loss": 2.2905, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.019217570350034317, | |
| "grad_norm": 4.628671169281006, | |
| "learning_rate": 0.00019791354817527755, | |
| "loss": 1.0654, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.01941366800666732, | |
| "grad_norm": 15.970938682556152, | |
| "learning_rate": 0.00019786602971838074, | |
| "loss": 2.9314, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.019609765663300322, | |
| "grad_norm": 5.425403594970703, | |
| "learning_rate": 0.00019781798206028239, | |
| "loss": 1.7236, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.019805863319933326, | |
| "grad_norm": 5.114929676055908, | |
| "learning_rate": 0.0001977694054607955, | |
| "loss": 2.1674, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.02000196097656633, | |
| "grad_norm": 9.564162254333496, | |
| "learning_rate": 0.0001977203001825935, | |
| "loss": 2.3091, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.020198058633199335, | |
| "grad_norm": 7.184217929840088, | |
| "learning_rate": 0.00019767066649120838, | |
| "loss": 2.5231, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.020394156289832335, | |
| "grad_norm": 8.926065444946289, | |
| "learning_rate": 0.00019762050465502965, | |
| "loss": 1.3343, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.02059025394646534, | |
| "grad_norm": 5.511591911315918, | |
| "learning_rate": 0.0001975698149453026, | |
| "loss": 1.7055, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.020786351603098344, | |
| "grad_norm": 8.730749130249023, | |
| "learning_rate": 0.00019751859763612704, | |
| "loss": 1.7862, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.020982449259731348, | |
| "grad_norm": 10.519305229187012, | |
| "learning_rate": 0.00019746685300445565, | |
| "loss": 1.9938, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.02117854691636435, | |
| "grad_norm": 6.307378768920898, | |
| "learning_rate": 0.00019741458133009258, | |
| "loss": 1.5184, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.021374644572997353, | |
| "grad_norm": 10.66562557220459, | |
| "learning_rate": 0.00019736178289569186, | |
| "loss": 2.0555, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.021570742229630357, | |
| "grad_norm": 6.235299587249756, | |
| "learning_rate": 0.0001973084579867561, | |
| "loss": 1.4547, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.021766839886263357, | |
| "grad_norm": 12.061420440673828, | |
| "learning_rate": 0.00019725460689163455, | |
| "loss": 1.9823, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.02196293754289636, | |
| "grad_norm": 8.891043663024902, | |
| "learning_rate": 0.0001972002299015219, | |
| "loss": 2.4757, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.022159035199529366, | |
| "grad_norm": 8.200632095336914, | |
| "learning_rate": 0.00019714532731045649, | |
| "loss": 1.4906, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.02235513285616237, | |
| "grad_norm": 8.106672286987305, | |
| "learning_rate": 0.00019708989941531887, | |
| "loss": 0.8863, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.02255123051279537, | |
| "grad_norm": 6.451066493988037, | |
| "learning_rate": 0.0001970339465158301, | |
| "loss": 2.1244, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.022747328169428375, | |
| "grad_norm": 6.3670220375061035, | |
| "learning_rate": 0.0001969774689145501, | |
| "loss": 1.3732, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.02294342582606138, | |
| "grad_norm": 6.401678562164307, | |
| "learning_rate": 0.0001969204669168761, | |
| "loss": 1.3963, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.023139523482694383, | |
| "grad_norm": 6.917253494262695, | |
| "learning_rate": 0.00019686294083104094, | |
| "loss": 1.3802, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.023335621139327384, | |
| "grad_norm": 4.852232456207275, | |
| "learning_rate": 0.00019680489096811149, | |
| "loss": 0.7538, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.023531718795960388, | |
| "grad_norm": 5.08411169052124, | |
| "learning_rate": 0.00019674631764198677, | |
| "loss": 1.5128, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.023727816452593392, | |
| "grad_norm": 13.64840316772461, | |
| "learning_rate": 0.00019668722116939649, | |
| "loss": 1.2718, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.023923914109226396, | |
| "grad_norm": 7.075209617614746, | |
| "learning_rate": 0.00019662760186989913, | |
| "loss": 2.4186, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.024120011765859397, | |
| "grad_norm": 12.186111450195312, | |
| "learning_rate": 0.00019656746006588044, | |
| "loss": 2.8495, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0243161094224924, | |
| "grad_norm": 13.221094131469727, | |
| "learning_rate": 0.00019650679608255138, | |
| "loss": 1.1832, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.024512207079125405, | |
| "grad_norm": 4.765728950500488, | |
| "learning_rate": 0.0001964456102479467, | |
| "loss": 1.1407, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.024708304735758406, | |
| "grad_norm": 4.750761032104492, | |
| "learning_rate": 0.00019638390289292295, | |
| "loss": 1.1096, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.02490440239239141, | |
| "grad_norm": 5.133049011230469, | |
| "learning_rate": 0.0001963216743511567, | |
| "loss": 2.6525, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.025100500049024414, | |
| "grad_norm": 14.541696548461914, | |
| "learning_rate": 0.0001962589249591429, | |
| "loss": 3.3739, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.02529659770565742, | |
| "grad_norm": 7.252123832702637, | |
| "learning_rate": 0.00019619565505619288, | |
| "loss": 2.0899, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.02549269536229042, | |
| "grad_norm": 7.14664888381958, | |
| "learning_rate": 0.00019613186498443257, | |
| "loss": 1.4538, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.025688793018923423, | |
| "grad_norm": 7.936334609985352, | |
| "learning_rate": 0.0001960675550888007, | |
| "loss": 1.351, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.025884890675556427, | |
| "grad_norm": 7.465827465057373, | |
| "learning_rate": 0.00019600272571704687, | |
| "loss": 1.0752, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.02608098833218943, | |
| "grad_norm": 12.85190486907959, | |
| "learning_rate": 0.00019593737721972977, | |
| "loss": 2.5674, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.026277085988822432, | |
| "grad_norm": 8.640382766723633, | |
| "learning_rate": 0.00019587150995021505, | |
| "loss": 2.5631, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.026473183645455436, | |
| "grad_norm": 5.070466995239258, | |
| "learning_rate": 0.00019580512426467376, | |
| "loss": 0.8935, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.02666928130208844, | |
| "grad_norm": 5.741654872894287, | |
| "learning_rate": 0.00019573822052208013, | |
| "loss": 2.1005, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.026865378958721445, | |
| "grad_norm": 8.615095138549805, | |
| "learning_rate": 0.00019567079908420972, | |
| "loss": 2.7478, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.027061476615354445, | |
| "grad_norm": 15.559739112854004, | |
| "learning_rate": 0.00019560286031563754, | |
| "loss": 1.8455, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.02725757427198745, | |
| "grad_norm": 5.683060169219971, | |
| "learning_rate": 0.000195534404583736, | |
| "loss": 1.6896, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.027453671928620454, | |
| "grad_norm": 5.791153430938721, | |
| "learning_rate": 0.00019546543225867292, | |
| "loss": 1.9291, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.027649769585253458, | |
| "grad_norm": 6.800760269165039, | |
| "learning_rate": 0.0001953959437134095, | |
| "loss": 1.789, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.02784586724188646, | |
| "grad_norm": 6.912458896636963, | |
| "learning_rate": 0.00019532593932369849, | |
| "loss": 2.2544, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.028041964898519463, | |
| "grad_norm": 6.928767681121826, | |
| "learning_rate": 0.00019525541946808188, | |
| "loss": 1.2531, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.028238062555152467, | |
| "grad_norm": 8.96179485321045, | |
| "learning_rate": 0.00019518438452788907, | |
| "loss": 2.3403, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.028434160211785468, | |
| "grad_norm": 7.004507064819336, | |
| "learning_rate": 0.00019511283488723473, | |
| "loss": 1.1211, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.028630257868418472, | |
| "grad_norm": 8.655360221862793, | |
| "learning_rate": 0.00019504077093301665, | |
| "loss": 1.6074, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.028826355525051476, | |
| "grad_norm": 7.188142776489258, | |
| "learning_rate": 0.00019496819305491383, | |
| "loss": 1.3564, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.02902245318168448, | |
| "grad_norm": 5.618204116821289, | |
| "learning_rate": 0.00019489510164538416, | |
| "loss": 2.5936, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.02921855083831748, | |
| "grad_norm": 6.578945159912109, | |
| "learning_rate": 0.00019482149709966246, | |
| "loss": 0.9577, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.029414648494950485, | |
| "grad_norm": 11.468759536743164, | |
| "learning_rate": 0.00019474737981575832, | |
| "loss": 2.1746, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02961074615158349, | |
| "grad_norm": 5.663360595703125, | |
| "learning_rate": 0.00019467275019445385, | |
| "loss": 1.2751, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.029806843808216493, | |
| "grad_norm": 6.297888278961182, | |
| "learning_rate": 0.00019459760863930155, | |
| "loss": 1.574, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.030002941464849494, | |
| "grad_norm": 8.760517120361328, | |
| "learning_rate": 0.00019452195555662224, | |
| "loss": 1.1148, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.030199039121482498, | |
| "grad_norm": 8.745152473449707, | |
| "learning_rate": 0.00019444579135550273, | |
| "loss": 1.4212, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.030395136778115502, | |
| "grad_norm": 7.594636917114258, | |
| "learning_rate": 0.00019436911644779366, | |
| "loss": 0.9161, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.030591234434748506, | |
| "grad_norm": 8.563089370727539, | |
| "learning_rate": 0.00019429193124810725, | |
| "loss": 1.5844, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.030787332091381507, | |
| "grad_norm": 16.4445743560791, | |
| "learning_rate": 0.00019421423617381508, | |
| "loss": 1.4798, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.03098342974801451, | |
| "grad_norm": 6.86074161529541, | |
| "learning_rate": 0.00019413603164504588, | |
| "loss": 2.3323, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.031179527404647515, | |
| "grad_norm": 11.293970108032227, | |
| "learning_rate": 0.0001940573180846831, | |
| "loss": 2.0304, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.03137562506128052, | |
| "grad_norm": 9.772544860839844, | |
| "learning_rate": 0.00019397809591836286, | |
| "loss": 3.1622, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.031571722717913524, | |
| "grad_norm": 8.489032745361328, | |
| "learning_rate": 0.00019389836557447143, | |
| "loss": 1.0113, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.03176782037454652, | |
| "grad_norm": 12.992219924926758, | |
| "learning_rate": 0.000193818127484143, | |
| "loss": 2.5478, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.031963918031179525, | |
| "grad_norm": 6.758875846862793, | |
| "learning_rate": 0.0001937373820812574, | |
| "loss": 1.8935, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.03216001568781253, | |
| "grad_norm": 11.88957405090332, | |
| "learning_rate": 0.0001936561298024377, | |
| "loss": 3.1899, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.03235611334444553, | |
| "grad_norm": 8.777862548828125, | |
| "learning_rate": 0.00019357437108704777, | |
| "loss": 2.7038, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.03255221100107854, | |
| "grad_norm": 5.135364532470703, | |
| "learning_rate": 0.0001934921063771901, | |
| "loss": 0.7644, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.03274830865771154, | |
| "grad_norm": 9.942065238952637, | |
| "learning_rate": 0.00019340933611770321, | |
| "loss": 1.4148, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.032944406314344546, | |
| "grad_norm": 7.881731033325195, | |
| "learning_rate": 0.0001933260607561594, | |
| "loss": 1.7402, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.03314050397097755, | |
| "grad_norm": 6.189711570739746, | |
| "learning_rate": 0.00019324228074286222, | |
| "loss": 1.642, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.03333660162761055, | |
| "grad_norm": 4.979406356811523, | |
| "learning_rate": 0.00019315799653084404, | |
| "loss": 2.425, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.03353269928424355, | |
| "grad_norm": 8.461871147155762, | |
| "learning_rate": 0.00019307320857586376, | |
| "loss": 2.6563, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.033728796940876556, | |
| "grad_norm": 8.992694854736328, | |
| "learning_rate": 0.00019298791733640406, | |
| "loss": 1.7962, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.03392489459750956, | |
| "grad_norm": 8.074629783630371, | |
| "learning_rate": 0.00019290212327366924, | |
| "loss": 1.5342, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.034120992254142564, | |
| "grad_norm": 6.5658111572265625, | |
| "learning_rate": 0.00019281582685158247, | |
| "loss": 0.7919, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.03431708991077557, | |
| "grad_norm": 9.946451187133789, | |
| "learning_rate": 0.00019272902853678336, | |
| "loss": 1.5664, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.03451318756740857, | |
| "grad_norm": 6.764862060546875, | |
| "learning_rate": 0.00019264172879862552, | |
| "loss": 2.1083, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.03470928522404157, | |
| "grad_norm": 11.195146560668945, | |
| "learning_rate": 0.000192553928109174, | |
| "loss": 1.3371, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.034905382880674574, | |
| "grad_norm": 6.352316856384277, | |
| "learning_rate": 0.00019246562694320255, | |
| "loss": 2.543, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.03510148053730758, | |
| "grad_norm": 4.030996799468994, | |
| "learning_rate": 0.00019237682577819137, | |
| "loss": 0.9273, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.03529757819394058, | |
| "grad_norm": 6.009339809417725, | |
| "learning_rate": 0.00019228752509432417, | |
| "loss": 2.1444, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.035493675850573586, | |
| "grad_norm": 14.911331176757812, | |
| "learning_rate": 0.00019219772537448597, | |
| "loss": 1.5989, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.03568977350720659, | |
| "grad_norm": 6.454592227935791, | |
| "learning_rate": 0.00019210742710426012, | |
| "loss": 1.0608, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.035885871163839594, | |
| "grad_norm": 11.354242324829102, | |
| "learning_rate": 0.00019201663077192586, | |
| "loss": 1.7558, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.0360819688204726, | |
| "grad_norm": 5.804329872131348, | |
| "learning_rate": 0.0001919253368684557, | |
| "loss": 1.411, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.036278066477105596, | |
| "grad_norm": 6.735583305358887, | |
| "learning_rate": 0.00019183354588751271, | |
| "loss": 2.4473, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.0364741641337386, | |
| "grad_norm": 9.169321060180664, | |
| "learning_rate": 0.00019174125832544786, | |
| "loss": 1.8947, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.036670261790371604, | |
| "grad_norm": 3.465175151824951, | |
| "learning_rate": 0.0001916484746812973, | |
| "loss": 1.1306, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.03686635944700461, | |
| "grad_norm": 10.71740436553955, | |
| "learning_rate": 0.0001915551954567797, | |
| "loss": 1.0832, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.03706245710363761, | |
| "grad_norm": 10.946560859680176, | |
| "learning_rate": 0.0001914614211562936, | |
| "loss": 1.9391, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.03725855476027062, | |
| "grad_norm": 6.847762584686279, | |
| "learning_rate": 0.0001913671522869145, | |
| "loss": 2.197, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.03745465241690362, | |
| "grad_norm": 13.330089569091797, | |
| "learning_rate": 0.00019127238935839235, | |
| "loss": 2.3539, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.03765075007353662, | |
| "grad_norm": 22.94158172607422, | |
| "learning_rate": 0.00019117713288314863, | |
| "loss": 2.868, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.03784684773016962, | |
| "grad_norm": 7.080881595611572, | |
| "learning_rate": 0.00019108138337627358, | |
| "loss": 1.7925, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.038042945386802626, | |
| "grad_norm": 4.726489067077637, | |
| "learning_rate": 0.00019098514135552357, | |
| "loss": 1.008, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.03823904304343563, | |
| "grad_norm": 5.7414870262146, | |
| "learning_rate": 0.00019088840734131807, | |
| "loss": 0.8934, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.038435140700068635, | |
| "grad_norm": 6.781312465667725, | |
| "learning_rate": 0.00019079118185673705, | |
| "loss": 1.6637, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.03863123835670164, | |
| "grad_norm": 12.264861106872559, | |
| "learning_rate": 0.00019069346542751803, | |
| "loss": 1.6055, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.03882733601333464, | |
| "grad_norm": 8.975689888000488, | |
| "learning_rate": 0.00019059525858205323, | |
| "loss": 2.6467, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.03902343366996765, | |
| "grad_norm": 9.027484893798828, | |
| "learning_rate": 0.0001904965618513868, | |
| "loss": 1.9813, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.039219531326600644, | |
| "grad_norm": 9.491011619567871, | |
| "learning_rate": 0.0001903973757692119, | |
| "loss": 1.4225, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.03941562898323365, | |
| "grad_norm": 5.202075481414795, | |
| "learning_rate": 0.00019029770087186773, | |
| "loss": 1.3524, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.03961172663986665, | |
| "grad_norm": 11.139556884765625, | |
| "learning_rate": 0.00019019753769833678, | |
| "loss": 2.0723, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.03980782429649966, | |
| "grad_norm": 11.383284568786621, | |
| "learning_rate": 0.0001900968867902419, | |
| "loss": 1.5843, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.04000392195313266, | |
| "grad_norm": 7.112687110900879, | |
| "learning_rate": 0.00018999574869184324, | |
| "loss": 1.3899, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.040200019609765665, | |
| "grad_norm": 13.50672721862793, | |
| "learning_rate": 0.00018989412395003537, | |
| "loss": 1.7484, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.04039611726639867, | |
| "grad_norm": 13.107057571411133, | |
| "learning_rate": 0.00018979201311434434, | |
| "loss": 1.6412, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.040592214923031666, | |
| "grad_norm": 11.221402168273926, | |
| "learning_rate": 0.0001896894167369248, | |
| "loss": 2.608, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.04078831257966467, | |
| "grad_norm": 4.945010662078857, | |
| "learning_rate": 0.0001895863353725568, | |
| "loss": 1.3582, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.040984410236297675, | |
| "grad_norm": 15.066801071166992, | |
| "learning_rate": 0.00018948276957864299, | |
| "loss": 1.7296, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.04118050789293068, | |
| "grad_norm": 7.497617244720459, | |
| "learning_rate": 0.0001893787199152055, | |
| "loss": 1.4961, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.04137660554956368, | |
| "grad_norm": 4.299473762512207, | |
| "learning_rate": 0.00018927418694488296, | |
| "loss": 1.7403, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.04157270320619669, | |
| "grad_norm": 6.988886833190918, | |
| "learning_rate": 0.00018916917123292738, | |
| "loss": 2.6546, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.04176880086282969, | |
| "grad_norm": 9.461721420288086, | |
| "learning_rate": 0.00018906367334720124, | |
| "loss": 2.067, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.041964898519462696, | |
| "grad_norm": 4.021213054656982, | |
| "learning_rate": 0.0001889576938581742, | |
| "loss": 1.0274, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.04216099617609569, | |
| "grad_norm": 7.254751205444336, | |
| "learning_rate": 0.00018885123333892026, | |
| "loss": 1.7091, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.0423570938327287, | |
| "grad_norm": 10.783350944519043, | |
| "learning_rate": 0.00018874429236511448, | |
| "loss": 1.3779, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.0425531914893617, | |
| "grad_norm": 16.38484764099121, | |
| "learning_rate": 0.00018863687151503, | |
| "loss": 2.8516, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.042749289145994705, | |
| "grad_norm": 5.848017692565918, | |
| "learning_rate": 0.00018852897136953473, | |
| "loss": 1.6814, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.04294538680262771, | |
| "grad_norm": 5.781267166137695, | |
| "learning_rate": 0.00018842059251208845, | |
| "loss": 2.1672, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.043141484459260714, | |
| "grad_norm": 6.244543552398682, | |
| "learning_rate": 0.00018831173552873946, | |
| "loss": 0.6934, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.04333758211589372, | |
| "grad_norm": 6.526815414428711, | |
| "learning_rate": 0.0001882024010081215, | |
| "loss": 1.6071, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.043533679772526715, | |
| "grad_norm": 6.064664363861084, | |
| "learning_rate": 0.00018809258954145052, | |
| "loss": 1.8964, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.04372977742915972, | |
| "grad_norm": 6.710343837738037, | |
| "learning_rate": 0.0001879823017225215, | |
| "loss": 2.063, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.04392587508579272, | |
| "grad_norm": 9.060622215270996, | |
| "learning_rate": 0.00018787153814770537, | |
| "loss": 1.568, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.04412197274242573, | |
| "grad_norm": 4.686062335968018, | |
| "learning_rate": 0.00018776029941594552, | |
| "loss": 1.1178, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.04431807039905873, | |
| "grad_norm": 6.257881164550781, | |
| "learning_rate": 0.00018764858612875472, | |
| "loss": 2.1195, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.044514168055691736, | |
| "grad_norm": 9.814234733581543, | |
| "learning_rate": 0.00018753639889021196, | |
| "loss": 1.1679, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.04471026571232474, | |
| "grad_norm": 13.946937561035156, | |
| "learning_rate": 0.00018742373830695898, | |
| "loss": 1.6899, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.044906363368957744, | |
| "grad_norm": 6.277964115142822, | |
| "learning_rate": 0.0001873106049881971, | |
| "loss": 1.5076, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.04510246102559074, | |
| "grad_norm": 14.092109680175781, | |
| "learning_rate": 0.00018719699954568398, | |
| "loss": 1.9726, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.045298558682223745, | |
| "grad_norm": 6.5708231925964355, | |
| "learning_rate": 0.00018708292259373015, | |
| "loss": 0.6623, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.04549465633885675, | |
| "grad_norm": 6.839974880218506, | |
| "learning_rate": 0.00018696837474919582, | |
| "loss": 1.8836, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.045690753995489754, | |
| "grad_norm": 17.919775009155273, | |
| "learning_rate": 0.00018685335663148753, | |
| "loss": 2.2343, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.04588685165212276, | |
| "grad_norm": 7.140262603759766, | |
| "learning_rate": 0.00018673786886255476, | |
| "loss": 1.9653, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.04608294930875576, | |
| "grad_norm": 11.049707412719727, | |
| "learning_rate": 0.00018662191206688658, | |
| "loss": 1.8658, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.046279046965388766, | |
| "grad_norm": 5.617869853973389, | |
| "learning_rate": 0.00018650548687150823, | |
| "loss": 1.862, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.04647514462202176, | |
| "grad_norm": 6.494004726409912, | |
| "learning_rate": 0.00018638859390597792, | |
| "loss": 3.787, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.04667124227865477, | |
| "grad_norm": 17.628082275390625, | |
| "learning_rate": 0.00018627123380238314, | |
| "loss": 1.9129, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.04686733993528777, | |
| "grad_norm": 5.634758949279785, | |
| "learning_rate": 0.0001861534071953374, | |
| "loss": 1.4702, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.047063437591920776, | |
| "grad_norm": 20.99329948425293, | |
| "learning_rate": 0.00018603511472197685, | |
| "loss": 2.2196, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.04725953524855378, | |
| "grad_norm": 6.650235652923584, | |
| "learning_rate": 0.00018591635702195673, | |
| "loss": 0.986, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.047455632905186784, | |
| "grad_norm": 4.837536334991455, | |
| "learning_rate": 0.00018579713473744795, | |
| "loss": 1.2033, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.04765173056181979, | |
| "grad_norm": 9.390655517578125, | |
| "learning_rate": 0.00018567744851313362, | |
| "loss": 1.7356, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.04784782821845279, | |
| "grad_norm": 8.881529808044434, | |
| "learning_rate": 0.0001855572989962056, | |
| "loss": 1.0063, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.04804392587508579, | |
| "grad_norm": 6.325936317443848, | |
| "learning_rate": 0.00018543668683636085, | |
| "loss": 1.5957, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.048240023531718794, | |
| "grad_norm": 10.866408348083496, | |
| "learning_rate": 0.0001853156126857981, | |
| "loss": 2.0472, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.0484361211883518, | |
| "grad_norm": 6.741912364959717, | |
| "learning_rate": 0.00018519407719921427, | |
| "loss": 1.6029, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.0486322188449848, | |
| "grad_norm": 5.979978561401367, | |
| "learning_rate": 0.00018507208103380092, | |
| "loss": 0.9353, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.048828316501617806, | |
| "grad_norm": 6.375998020172119, | |
| "learning_rate": 0.00018494962484924058, | |
| "loss": 2.5973, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.04902441415825081, | |
| "grad_norm": 8.528829574584961, | |
| "learning_rate": 0.00018482670930770342, | |
| "loss": 1.1618, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.049220511814883815, | |
| "grad_norm": 7.874660491943359, | |
| "learning_rate": 0.0001847033350738435, | |
| "loss": 1.3043, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.04941660947151681, | |
| "grad_norm": 4.421433925628662, | |
| "learning_rate": 0.00018457950281479513, | |
| "loss": 2.0768, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.049612707128149816, | |
| "grad_norm": 8.347919464111328, | |
| "learning_rate": 0.00018445521320016944, | |
| "loss": 1.0983, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.04980880478478282, | |
| "grad_norm": 6.713651657104492, | |
| "learning_rate": 0.00018433046690205068, | |
| "loss": 0.9891, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.050004902441415824, | |
| "grad_norm": 12.359843254089355, | |
| "learning_rate": 0.0001842052645949925, | |
| "loss": 2.5271, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.05020100009804883, | |
| "grad_norm": 4.7271199226379395, | |
| "learning_rate": 0.00018407960695601442, | |
| "loss": 1.394, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.05039709775468183, | |
| "grad_norm": 5.810708522796631, | |
| "learning_rate": 0.0001839534946645981, | |
| "loss": 1.7354, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.05059319541131484, | |
| "grad_norm": 28.575908660888672, | |
| "learning_rate": 0.00018382692840268367, | |
| "loss": 3.4793, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.05078929306794784, | |
| "grad_norm": 5.775376319885254, | |
| "learning_rate": 0.00018369990885466617, | |
| "loss": 1.4695, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.05098539072458084, | |
| "grad_norm": 7.531515121459961, | |
| "learning_rate": 0.0001835724367073916, | |
| "loss": 2.0772, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.05118148838121384, | |
| "grad_norm": 5.099686145782471, | |
| "learning_rate": 0.00018344451265015348, | |
| "loss": 3.2733, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.05137758603784685, | |
| "grad_norm": 5.558218479156494, | |
| "learning_rate": 0.00018331613737468887, | |
| "loss": 1.7578, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.05157368369447985, | |
| "grad_norm": 7.837512016296387, | |
| "learning_rate": 0.00018318731157517478, | |
| "loss": 2.7413, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.051769781351112855, | |
| "grad_norm": 9.013374328613281, | |
| "learning_rate": 0.00018305803594822448, | |
| "loss": 1.7722, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.05196587900774586, | |
| "grad_norm": 7.108828067779541, | |
| "learning_rate": 0.00018292831119288348, | |
| "loss": 1.31, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.05216197666437886, | |
| "grad_norm": 6.387202739715576, | |
| "learning_rate": 0.0001827981380106261, | |
| "loss": 1.0588, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.05235807432101187, | |
| "grad_norm": 13.306784629821777, | |
| "learning_rate": 0.00018266751710535131, | |
| "loss": 2.6092, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.052554171977644865, | |
| "grad_norm": 7.774232864379883, | |
| "learning_rate": 0.00018253644918337915, | |
| "loss": 1.2318, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.05275026963427787, | |
| "grad_norm": 5.396208763122559, | |
| "learning_rate": 0.00018240493495344694, | |
| "loss": 1.2144, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.05294636729091087, | |
| "grad_norm": 13.393839836120605, | |
| "learning_rate": 0.0001822729751267053, | |
| "loss": 2.4278, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.05314246494754388, | |
| "grad_norm": 14.873156547546387, | |
| "learning_rate": 0.00018214057041671434, | |
| "loss": 2.4015, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.05333856260417688, | |
| "grad_norm": 7.224187850952148, | |
| "learning_rate": 0.00018200772153943988, | |
| "loss": 1.578, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.053534660260809885, | |
| "grad_norm": 16.03417205810547, | |
| "learning_rate": 0.00018187442921324958, | |
| "loss": 2.4634, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.05373075791744289, | |
| "grad_norm": 6.284254550933838, | |
| "learning_rate": 0.00018174069415890888, | |
| "loss": 1.1995, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.05392685557407589, | |
| "grad_norm": 5.986271858215332, | |
| "learning_rate": 0.00018160651709957736, | |
| "loss": 1.8718, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.05412295323070889, | |
| "grad_norm": 13.663339614868164, | |
| "learning_rate": 0.00018147189876080463, | |
| "loss": 1.8664, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.054319050887341895, | |
| "grad_norm": 8.843401908874512, | |
| "learning_rate": 0.0001813368398705265, | |
| "loss": 4.1059, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.0545151485439749, | |
| "grad_norm": 6.658843994140625, | |
| "learning_rate": 0.00018120134115906096, | |
| "loss": 0.8269, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.0547112462006079, | |
| "grad_norm": 14.153685569763184, | |
| "learning_rate": 0.0001810654033591044, | |
| "loss": 1.9537, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.05490734385724091, | |
| "grad_norm": 5.585552215576172, | |
| "learning_rate": 0.00018092902720572745, | |
| "loss": 1.2531, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.05510344151387391, | |
| "grad_norm": 6.414156436920166, | |
| "learning_rate": 0.00018079221343637113, | |
| "loss": 0.9456, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.055299539170506916, | |
| "grad_norm": 6.674518585205078, | |
| "learning_rate": 0.00018065496279084283, | |
| "loss": 1.1899, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.05549563682713991, | |
| "grad_norm": 16.878883361816406, | |
| "learning_rate": 0.00018051727601131227, | |
| "loss": 0.8761, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.05569173448377292, | |
| "grad_norm": 31.99020004272461, | |
| "learning_rate": 0.0001803791538423076, | |
| "loss": 1.7235, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.05588783214040592, | |
| "grad_norm": 4.821253776550293, | |
| "learning_rate": 0.0001802405970307112, | |
| "loss": 1.2902, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.056083929797038926, | |
| "grad_norm": 7.2293267250061035, | |
| "learning_rate": 0.00018010160632575577, | |
| "loss": 1.0864, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.05628002745367193, | |
| "grad_norm": 8.262434959411621, | |
| "learning_rate": 0.00017996218247902035, | |
| "loss": 1.6246, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.056476125110304934, | |
| "grad_norm": 7.845222473144531, | |
| "learning_rate": 0.00017982232624442595, | |
| "loss": 1.7249, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.05667222276693794, | |
| "grad_norm": 9.781503677368164, | |
| "learning_rate": 0.0001796820383782319, | |
| "loss": 1.9911, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.056868320423570935, | |
| "grad_norm": 10.7435884475708, | |
| "learning_rate": 0.00017954131963903133, | |
| "loss": 1.4689, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.05706441808020394, | |
| "grad_norm": 8.709835052490234, | |
| "learning_rate": 0.00017940017078774747, | |
| "loss": 2.3939, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.057260515736836944, | |
| "grad_norm": 6.140249252319336, | |
| "learning_rate": 0.00017925859258762915, | |
| "loss": 2.0753, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.05745661339346995, | |
| "grad_norm": 9.725993156433105, | |
| "learning_rate": 0.00017911658580424704, | |
| "loss": 1.5315, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.05765271105010295, | |
| "grad_norm": 20.413393020629883, | |
| "learning_rate": 0.00017897415120548917, | |
| "loss": 2.4083, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.057848808706735956, | |
| "grad_norm": 7.881735324859619, | |
| "learning_rate": 0.00017883128956155706, | |
| "loss": 3.1099, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.05804490636336896, | |
| "grad_norm": 8.856921195983887, | |
| "learning_rate": 0.0001786880016449614, | |
| "loss": 0.9566, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.058241004020001964, | |
| "grad_norm": 25.80061912536621, | |
| "learning_rate": 0.0001785442882305179, | |
| "loss": 2.6205, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.05843710167663496, | |
| "grad_norm": 13.293787956237793, | |
| "learning_rate": 0.00017840015009534308, | |
| "loss": 1.7317, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.058633199333267966, | |
| "grad_norm": 11.313176155090332, | |
| "learning_rate": 0.00017825558801885016, | |
| "loss": 1.5122, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.05882929698990097, | |
| "grad_norm": 9.636054039001465, | |
| "learning_rate": 0.00017811060278274474, | |
| "loss": 2.1976, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.059025394646533974, | |
| "grad_norm": 11.10985279083252, | |
| "learning_rate": 0.00017796519517102066, | |
| "loss": 2.5926, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.05922149230316698, | |
| "grad_norm": 6.27458381652832, | |
| "learning_rate": 0.00017781936596995563, | |
| "loss": 2.5326, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.05941758995979998, | |
| "grad_norm": 10.134029388427734, | |
| "learning_rate": 0.00017767311596810715, | |
| "loss": 1.4142, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.059613687616432987, | |
| "grad_norm": 10.580477714538574, | |
| "learning_rate": 0.0001775264459563081, | |
| "loss": 2.3389, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.059809785273065984, | |
| "grad_norm": 7.545969009399414, | |
| "learning_rate": 0.00017737935672766257, | |
| "loss": 2.3728, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.06000588292969899, | |
| "grad_norm": 6.5107622146606445, | |
| "learning_rate": 0.00017723184907754154, | |
| "loss": 1.0087, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.06020198058633199, | |
| "grad_norm": 10.663747787475586, | |
| "learning_rate": 0.00017708392380357845, | |
| "loss": 2.2001, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.060398078242964996, | |
| "grad_norm": 10.211285591125488, | |
| "learning_rate": 0.0001769355817056651, | |
| "loss": 2.2537, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.060594175899598, | |
| "grad_norm": 8.512606620788574, | |
| "learning_rate": 0.00017678682358594728, | |
| "loss": 1.0687, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.060790273556231005, | |
| "grad_norm": 10.64330768585205, | |
| "learning_rate": 0.0001766376502488202, | |
| "loss": 2.4672, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.06098637121286401, | |
| "grad_norm": 9.972363471984863, | |
| "learning_rate": 0.0001764880625009245, | |
| "loss": 2.2277, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.06118246886949701, | |
| "grad_norm": 10.01198959350586, | |
| "learning_rate": 0.0001763380611511416, | |
| "loss": 1.906, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.06137856652613001, | |
| "grad_norm": 11.719582557678223, | |
| "learning_rate": 0.00017618764701058949, | |
| "loss": 2.4849, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.061574664182763014, | |
| "grad_norm": 9.08901309967041, | |
| "learning_rate": 0.0001760368208926182, | |
| "loss": 1.4476, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.06177076183939602, | |
| "grad_norm": 11.74946403503418, | |
| "learning_rate": 0.00017588558361280557, | |
| "loss": 1.3607, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.06196685949602902, | |
| "grad_norm": 6.549604892730713, | |
| "learning_rate": 0.00017573393598895276, | |
| "loss": 2.066, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.06216295715266203, | |
| "grad_norm": 5.160909652709961, | |
| "learning_rate": 0.00017558187884107978, | |
| "loss": 0.9799, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.06235905480929503, | |
| "grad_norm": 11.00747299194336, | |
| "learning_rate": 0.00017542941299142112, | |
| "loss": 2.3193, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.06255515246592804, | |
| "grad_norm": 4.467062473297119, | |
| "learning_rate": 0.00017527653926442135, | |
| "loss": 1.9208, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.06275125012256104, | |
| "grad_norm": 11.554845809936523, | |
| "learning_rate": 0.00017512325848673043, | |
| "loss": 1.5197, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.06294734777919404, | |
| "grad_norm": 6.626307964324951, | |
| "learning_rate": 0.0001749695714871996, | |
| "loss": 2.0222, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.06314344543582705, | |
| "grad_norm": 5.357675075531006, | |
| "learning_rate": 0.00017481547909687658, | |
| "loss": 2.3264, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.06333954309246005, | |
| "grad_norm": 9.451851844787598, | |
| "learning_rate": 0.00017466098214900124, | |
| "loss": 1.5699, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.06353564074909304, | |
| "grad_norm": 5.937804698944092, | |
| "learning_rate": 0.00017450608147900106, | |
| "loss": 1.832, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.06373173840572605, | |
| "grad_norm": 10.744606971740723, | |
| "learning_rate": 0.00017435077792448664, | |
| "loss": 1.4665, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.06392783606235905, | |
| "grad_norm": 9.900199890136719, | |
| "learning_rate": 0.0001741950723252471, | |
| "loss": 1.4785, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.06412393371899205, | |
| "grad_norm": 13.663531303405762, | |
| "learning_rate": 0.00017403896552324553, | |
| "loss": 0.9564, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.06432003137562506, | |
| "grad_norm": 4.708930015563965, | |
| "learning_rate": 0.00017388245836261464, | |
| "loss": 0.8293, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.06451612903225806, | |
| "grad_norm": 6.656357288360596, | |
| "learning_rate": 0.00017372555168965184, | |
| "loss": 2.2177, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.06471222668889107, | |
| "grad_norm": 6.349363803863525, | |
| "learning_rate": 0.00017356824635281502, | |
| "loss": 3.3319, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.06490832434552407, | |
| "grad_norm": 10.561055183410645, | |
| "learning_rate": 0.00017341054320271776, | |
| "loss": 1.919, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.06510442200215708, | |
| "grad_norm": 10.805617332458496, | |
| "learning_rate": 0.00017325244309212475, | |
| "loss": 2.9898, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.06530051965879008, | |
| "grad_norm": 9.67844009399414, | |
| "learning_rate": 0.0001730939468759472, | |
| "loss": 3.0727, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.06549661731542308, | |
| "grad_norm": 9.487295150756836, | |
| "learning_rate": 0.00017293505541123833, | |
| "loss": 1.2741, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.06569271497205609, | |
| "grad_norm": 5.027645111083984, | |
| "learning_rate": 0.00017277576955718847, | |
| "loss": 1.6962, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.06588881262868909, | |
| "grad_norm": 6.2696685791015625, | |
| "learning_rate": 0.0001726160901751207, | |
| "loss": 1.4388, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.0660849102853221, | |
| "grad_norm": 9.509906768798828, | |
| "learning_rate": 0.000172456018128486, | |
| "loss": 2.3173, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.0662810079419551, | |
| "grad_norm": 5.882188320159912, | |
| "learning_rate": 0.00017229555428285864, | |
| "loss": 1.1832, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.06647710559858809, | |
| "grad_norm": 8.857211112976074, | |
| "learning_rate": 0.00017213469950593156, | |
| "loss": 2.484, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.0666732032552211, | |
| "grad_norm": 10.660289764404297, | |
| "learning_rate": 0.00017197345466751158, | |
| "loss": 2.5169, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.0668693009118541, | |
| "grad_norm": 6.141770839691162, | |
| "learning_rate": 0.00017181182063951474, | |
| "loss": 1.601, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.0668693009118541, | |
| "eval_loss": 0.4308730363845825, | |
| "eval_runtime": 77.758, | |
| "eval_samples_per_second": 27.624, | |
| "eval_steps_per_second": 13.812, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.0670653985684871, | |
| "grad_norm": 9.668785095214844, | |
| "learning_rate": 0.00017164979829596165, | |
| "loss": 3.3758, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.06726149622512011, | |
| "grad_norm": 5.6129255294799805, | |
| "learning_rate": 0.00017148738851297256, | |
| "loss": 1.3369, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.06745759388175311, | |
| "grad_norm": 8.042420387268066, | |
| "learning_rate": 0.0001713245921687629, | |
| "loss": 1.918, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.06765369153838612, | |
| "grad_norm": 6.536214351654053, | |
| "learning_rate": 0.00017116141014363837, | |
| "loss": 0.97, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.06784978919501912, | |
| "grad_norm": 10.576717376708984, | |
| "learning_rate": 0.0001709978433199901, | |
| "loss": 2.0315, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.06804588685165212, | |
| "grad_norm": 13.645353317260742, | |
| "learning_rate": 0.00017083389258229013, | |
| "loss": 2.3226, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.06824198450828513, | |
| "grad_norm": 10.135254859924316, | |
| "learning_rate": 0.00017066955881708636, | |
| "loss": 2.4686, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.06843808216491813, | |
| "grad_norm": 5.7091569900512695, | |
| "learning_rate": 0.0001705048429129979, | |
| "loss": 1.633, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.06863417982155114, | |
| "grad_norm": 7.3178935050964355, | |
| "learning_rate": 0.0001703397457607103, | |
| "loss": 1.5411, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.06883027747818414, | |
| "grad_norm": 10.689863204956055, | |
| "learning_rate": 0.0001701742682529706, | |
| "loss": 2.8238, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.06902637513481714, | |
| "grad_norm": 4.083006381988525, | |
| "learning_rate": 0.00017000841128458265, | |
| "loss": 1.7831, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.06922247279145015, | |
| "grad_norm": 4.592381477355957, | |
| "learning_rate": 0.0001698421757524021, | |
| "loss": 1.1135, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.06941857044808314, | |
| "grad_norm": 9.44848346710205, | |
| "learning_rate": 0.00016967556255533174, | |
| "loss": 1.6301, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.06961466810471614, | |
| "grad_norm": 7.35884428024292, | |
| "learning_rate": 0.0001695085725943165, | |
| "loss": 1.6776, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.06981076576134915, | |
| "grad_norm": 7.0129475593566895, | |
| "learning_rate": 0.00016934120677233863, | |
| "loss": 1.1665, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.07000686341798215, | |
| "grad_norm": 11.612259864807129, | |
| "learning_rate": 0.00016917346599441276, | |
| "loss": 0.9044, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.07020296107461516, | |
| "grad_norm": 5.861693382263184, | |
| "learning_rate": 0.00016900535116758125, | |
| "loss": 1.5738, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.07039905873124816, | |
| "grad_norm": 4.579985618591309, | |
| "learning_rate": 0.00016883686320090884, | |
| "loss": 1.6305, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.07059515638788116, | |
| "grad_norm": 6.142041206359863, | |
| "learning_rate": 0.00016866800300547813, | |
| "loss": 0.6206, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.07079125404451417, | |
| "grad_norm": 3.1343085765838623, | |
| "learning_rate": 0.0001684987714943845, | |
| "loss": 0.7649, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.07098735170114717, | |
| "grad_norm": 5.473539352416992, | |
| "learning_rate": 0.00016832916958273118, | |
| "loss": 1.4839, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.07118344935778018, | |
| "grad_norm": 8.52957534790039, | |
| "learning_rate": 0.00016815919818762427, | |
| "loss": 2.5437, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.07137954701441318, | |
| "grad_norm": 6.140387058258057, | |
| "learning_rate": 0.00016798885822816786, | |
| "loss": 1.2282, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.07157564467104618, | |
| "grad_norm": 9.70067024230957, | |
| "learning_rate": 0.000167818150625459, | |
| "loss": 2.3286, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.07177174232767919, | |
| "grad_norm": 6.14666223526001, | |
| "learning_rate": 0.0001676470763025827, | |
| "loss": 1.3825, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.07196783998431219, | |
| "grad_norm": 10.377222061157227, | |
| "learning_rate": 0.0001674756361846071, | |
| "loss": 1.9278, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.0721639376409452, | |
| "grad_norm": 7.09204626083374, | |
| "learning_rate": 0.00016730383119857817, | |
| "loss": 1.2413, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.07236003529757819, | |
| "grad_norm": 12.448009490966797, | |
| "learning_rate": 0.00016713166227351497, | |
| "loss": 1.7415, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.07255613295421119, | |
| "grad_norm": 4.5050811767578125, | |
| "learning_rate": 0.00016695913034040454, | |
| "loss": 1.7608, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.0727522306108442, | |
| "grad_norm": 8.113517761230469, | |
| "learning_rate": 0.00016678623633219677, | |
| "loss": 1.8336, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.0729483282674772, | |
| "grad_norm": 7.662284851074219, | |
| "learning_rate": 0.00016661298118379948, | |
| "loss": 2.5928, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.0731444259241102, | |
| "grad_norm": 7.647708892822266, | |
| "learning_rate": 0.00016643936583207337, | |
| "loss": 3.006, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.07334052358074321, | |
| "grad_norm": 7.049048900604248, | |
| "learning_rate": 0.00016626539121582685, | |
| "loss": 1.9476, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.07353662123737621, | |
| "grad_norm": 4.551039218902588, | |
| "learning_rate": 0.000166091058275811, | |
| "loss": 2.2753, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.07373271889400922, | |
| "grad_norm": 8.680680274963379, | |
| "learning_rate": 0.00016591636795471455, | |
| "loss": 2.0624, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.07392881655064222, | |
| "grad_norm": 10.324525833129883, | |
| "learning_rate": 0.0001657413211971587, | |
| "loss": 1.6293, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.07412491420727522, | |
| "grad_norm": 5.324687957763672, | |
| "learning_rate": 0.0001655659189496921, | |
| "loss": 1.4987, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.07432101186390823, | |
| "grad_norm": 11.023552894592285, | |
| "learning_rate": 0.00016539016216078557, | |
| "loss": 1.9108, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.07451710952054123, | |
| "grad_norm": 6.6586594581604, | |
| "learning_rate": 0.00016521405178082722, | |
| "loss": 2.4198, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.07471320717717424, | |
| "grad_norm": 9.818254470825195, | |
| "learning_rate": 0.0001650375887621171, | |
| "loss": 2.162, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.07490930483380724, | |
| "grad_norm": 6.547168254852295, | |
| "learning_rate": 0.000164860774058862, | |
| "loss": 1.8307, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.07510540249044025, | |
| "grad_norm": 7.256173610687256, | |
| "learning_rate": 0.00016468360862717066, | |
| "loss": 0.9432, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.07530150014707324, | |
| "grad_norm": 10.348817825317383, | |
| "learning_rate": 0.00016450609342504813, | |
| "loss": 1.7824, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.07549759780370624, | |
| "grad_norm": 14.728826522827148, | |
| "learning_rate": 0.00016432822941239096, | |
| "loss": 3.1737, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.07569369546033924, | |
| "grad_norm": 7.894331932067871, | |
| "learning_rate": 0.00016415001755098175, | |
| "loss": 2.0515, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.07588979311697225, | |
| "grad_norm": 7.474819183349609, | |
| "learning_rate": 0.00016397145880448416, | |
| "loss": 3.1276, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.07608589077360525, | |
| "grad_norm": 6.2442626953125, | |
| "learning_rate": 0.00016379255413843754, | |
| "loss": 0.9294, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.07628198843023826, | |
| "grad_norm": 8.229086875915527, | |
| "learning_rate": 0.0001636133045202517, | |
| "loss": 2.122, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.07647808608687126, | |
| "grad_norm": 5.9484100341796875, | |
| "learning_rate": 0.0001634337109192019, | |
| "loss": 1.3164, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.07667418374350427, | |
| "grad_norm": 6.490477561950684, | |
| "learning_rate": 0.0001632537743064233, | |
| "loss": 1.1598, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.07687028140013727, | |
| "grad_norm": 36.34891128540039, | |
| "learning_rate": 0.000163073495654906, | |
| "loss": 2.4203, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.07706637905677027, | |
| "grad_norm": 5.949688911437988, | |
| "learning_rate": 0.00016289287593948952, | |
| "loss": 2.0893, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.07726247671340328, | |
| "grad_norm": 5.93861198425293, | |
| "learning_rate": 0.00016271191613685776, | |
| "loss": 1.829, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.07745857437003628, | |
| "grad_norm": 8.41703987121582, | |
| "learning_rate": 0.00016253061722553355, | |
| "loss": 2.1544, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.07765467202666929, | |
| "grad_norm": 8.92647647857666, | |
| "learning_rate": 0.00016234898018587337, | |
| "loss": 1.3316, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.07785076968330229, | |
| "grad_norm": 4.4784698486328125, | |
| "learning_rate": 0.0001621670060000622, | |
| "loss": 1.0305, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.0780468673399353, | |
| "grad_norm": 10.080864906311035, | |
| "learning_rate": 0.00016198469565210805, | |
| "loss": 2.1012, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.07824296499656828, | |
| "grad_norm": 8.505258560180664, | |
| "learning_rate": 0.0001618020501278367, | |
| "loss": 2.2348, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.07843906265320129, | |
| "grad_norm": 6.3116559982299805, | |
| "learning_rate": 0.00016161907041488635, | |
| "loss": 1.3726, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.07863516030983429, | |
| "grad_norm": 7.111060619354248, | |
| "learning_rate": 0.00016143575750270233, | |
| "loss": 1.5926, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.0788312579664673, | |
| "grad_norm": 5.481679916381836, | |
| "learning_rate": 0.0001612521123825317, | |
| "loss": 0.8062, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.0790273556231003, | |
| "grad_norm": 9.399847984313965, | |
| "learning_rate": 0.00016106813604741782, | |
| "loss": 1.7107, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.0792234532797333, | |
| "grad_norm": 7.784383296966553, | |
| "learning_rate": 0.00016088382949219524, | |
| "loss": 1.4449, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.07941955093636631, | |
| "grad_norm": 10.075730323791504, | |
| "learning_rate": 0.000160699193713484, | |
| "loss": 2.3277, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.07961564859299931, | |
| "grad_norm": 4.376564979553223, | |
| "learning_rate": 0.00016051422970968438, | |
| "loss": 1.3489, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.07981174624963232, | |
| "grad_norm": 28.959577560424805, | |
| "learning_rate": 0.00016032893848097165, | |
| "loss": 0.9703, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.08000784390626532, | |
| "grad_norm": 9.604166030883789, | |
| "learning_rate": 0.00016014332102929027, | |
| "loss": 2.195, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.08020394156289833, | |
| "grad_norm": 5.949897289276123, | |
| "learning_rate": 0.00015995737835834906, | |
| "loss": 0.8828, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.08040003921953133, | |
| "grad_norm": 9.921558380126953, | |
| "learning_rate": 0.00015977111147361507, | |
| "loss": 2.1569, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.08059613687616433, | |
| "grad_norm": 9.299111366271973, | |
| "learning_rate": 0.00015958452138230877, | |
| "loss": 1.4095, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.08079223453279734, | |
| "grad_norm": 8.806063652038574, | |
| "learning_rate": 0.00015939760909339823, | |
| "loss": 1.7785, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.08098833218943034, | |
| "grad_norm": 5.1091108322143555, | |
| "learning_rate": 0.00015921037561759377, | |
| "loss": 1.8415, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.08118442984606333, | |
| "grad_norm": 12.836872100830078, | |
| "learning_rate": 0.0001590228219673425, | |
| "loss": 1.8644, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.08138052750269634, | |
| "grad_norm": 6.567033767700195, | |
| "learning_rate": 0.00015883494915682289, | |
| "loss": 1.9557, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.08157662515932934, | |
| "grad_norm": 8.996990203857422, | |
| "learning_rate": 0.00015864675820193922, | |
| "loss": 1.888, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.08177272281596235, | |
| "grad_norm": 7.494692802429199, | |
| "learning_rate": 0.00015845825012031605, | |
| "loss": 2.4514, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.08196882047259535, | |
| "grad_norm": 5.421525478363037, | |
| "learning_rate": 0.0001582694259312928, | |
| "loss": 1.0314, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.08216491812922835, | |
| "grad_norm": 25.4990291595459, | |
| "learning_rate": 0.0001580802866559183, | |
| "loss": 0.9981, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.08236101578586136, | |
| "grad_norm": 7.055636405944824, | |
| "learning_rate": 0.00015789083331694506, | |
| "loss": 1.718, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.08255711344249436, | |
| "grad_norm": 8.986098289489746, | |
| "learning_rate": 0.00015770106693882387, | |
| "loss": 1.3962, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.08275321109912737, | |
| "grad_norm": 4.7463788986206055, | |
| "learning_rate": 0.0001575109885476983, | |
| "loss": 1.4039, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.08294930875576037, | |
| "grad_norm": 6.555906295776367, | |
| "learning_rate": 0.00015732059917139912, | |
| "loss": 1.0908, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.08314540641239337, | |
| "grad_norm": 7.233346462249756, | |
| "learning_rate": 0.00015712989983943863, | |
| "loss": 1.9737, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.08334150406902638, | |
| "grad_norm": 4.797744274139404, | |
| "learning_rate": 0.0001569388915830053, | |
| "loss": 1.2191, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.08353760172565938, | |
| "grad_norm": 6.303066253662109, | |
| "learning_rate": 0.000156747575434958, | |
| "loss": 1.8937, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.08373369938229239, | |
| "grad_norm": 12.17996597290039, | |
| "learning_rate": 0.00015655595242982048, | |
| "loss": 4.1888, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.08392979703892539, | |
| "grad_norm": 7.345923900604248, | |
| "learning_rate": 0.00015636402360377587, | |
| "loss": 1.0776, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.08412589469555838, | |
| "grad_norm": 7.688333034515381, | |
| "learning_rate": 0.0001561717899946609, | |
| "loss": 0.9205, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.08432199235219139, | |
| "grad_norm": 5.487595558166504, | |
| "learning_rate": 0.00015597925264196049, | |
| "loss": 1.4256, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.08451809000882439, | |
| "grad_norm": 7.871713161468506, | |
| "learning_rate": 0.0001557864125868019, | |
| "loss": 2.3536, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.0847141876654574, | |
| "grad_norm": 5.0042033195495605, | |
| "learning_rate": 0.00015559327087194942, | |
| "loss": 3.0954, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.0849102853220904, | |
| "grad_norm": 6.770071506500244, | |
| "learning_rate": 0.0001553998285417983, | |
| "loss": 1.5016, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.0851063829787234, | |
| "grad_norm": 6.884315490722656, | |
| "learning_rate": 0.00015520608664236949, | |
| "loss": 0.9995, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.0853024806353564, | |
| "grad_norm": 7.716054439544678, | |
| "learning_rate": 0.00015501204622130377, | |
| "loss": 1.525, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.08549857829198941, | |
| "grad_norm": 9.22014045715332, | |
| "learning_rate": 0.0001548177083278562, | |
| "loss": 1.7368, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.08569467594862241, | |
| "grad_norm": 11.081851959228516, | |
| "learning_rate": 0.0001546230740128904, | |
| "loss": 2.5292, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.08589077360525542, | |
| "grad_norm": 3.8471319675445557, | |
| "learning_rate": 0.0001544281443288728, | |
| "loss": 1.8473, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.08608687126188842, | |
| "grad_norm": 7.554915904998779, | |
| "learning_rate": 0.00015423292032986695, | |
| "loss": 1.4886, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.08628296891852143, | |
| "grad_norm": 5.459715843200684, | |
| "learning_rate": 0.00015403740307152805, | |
| "loss": 2.7721, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.08647906657515443, | |
| "grad_norm": 5.094114780426025, | |
| "learning_rate": 0.00015384159361109698, | |
| "loss": 0.6022, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.08667516423178744, | |
| "grad_norm": 4.084317684173584, | |
| "learning_rate": 0.00015364549300739467, | |
| "loss": 0.8871, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.08687126188842044, | |
| "grad_norm": 3.810145378112793, | |
| "learning_rate": 0.0001534491023208164, | |
| "loss": 2.2058, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.08706735954505343, | |
| "grad_norm": 9.405007362365723, | |
| "learning_rate": 0.000153252422613326, | |
| "loss": 2.414, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.08726345720168643, | |
| "grad_norm": 8.687947273254395, | |
| "learning_rate": 0.00015305545494845023, | |
| "loss": 0.8809, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.08745955485831944, | |
| "grad_norm": 6.396414756774902, | |
| "learning_rate": 0.00015285820039127293, | |
| "loss": 1.488, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.08765565251495244, | |
| "grad_norm": 4.447728157043457, | |
| "learning_rate": 0.00015266066000842917, | |
| "loss": 1.1921, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.08785175017158545, | |
| "grad_norm": 9.093881607055664, | |
| "learning_rate": 0.00015246283486809977, | |
| "loss": 1.6693, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.08804784782821845, | |
| "grad_norm": 8.160088539123535, | |
| "learning_rate": 0.0001522647260400053, | |
| "loss": 2.9597, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.08824394548485145, | |
| "grad_norm": 4.759898662567139, | |
| "learning_rate": 0.00015206633459540023, | |
| "loss": 1.121, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.08844004314148446, | |
| "grad_norm": 6.537527561187744, | |
| "learning_rate": 0.0001518676616070674, | |
| "loss": 1.1842, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.08863614079811746, | |
| "grad_norm": 3.6689343452453613, | |
| "learning_rate": 0.0001516687081493119, | |
| "loss": 1.3464, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.08883223845475047, | |
| "grad_norm": 13.292744636535645, | |
| "learning_rate": 0.00015146947529795567, | |
| "loss": 2.8284, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.08902833611138347, | |
| "grad_norm": 3.7040257453918457, | |
| "learning_rate": 0.0001512699641303312, | |
| "loss": 1.0223, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.08922443376801648, | |
| "grad_norm": 8.529545783996582, | |
| "learning_rate": 0.00015107017572527616, | |
| "loss": 1.9594, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.08942053142464948, | |
| "grad_norm": 11.990189552307129, | |
| "learning_rate": 0.00015087011116312718, | |
| "loss": 1.2631, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.08961662908128248, | |
| "grad_norm": 5.5952534675598145, | |
| "learning_rate": 0.00015066977152571428, | |
| "loss": 0.9935, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.08981272673791549, | |
| "grad_norm": 6.168736457824707, | |
| "learning_rate": 0.0001504691578963549, | |
| "loss": 0.9699, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.09000882439454848, | |
| "grad_norm": 6.674266338348389, | |
| "learning_rate": 0.0001502682713598481, | |
| "loss": 1.3238, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.09020492205118148, | |
| "grad_norm": 7.915739059448242, | |
| "learning_rate": 0.00015006711300246853, | |
| "loss": 1.4752, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.09040101970781449, | |
| "grad_norm": 9.858468055725098, | |
| "learning_rate": 0.00014986568391196092, | |
| "loss": 1.7595, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.09059711736444749, | |
| "grad_norm": 7.993055820465088, | |
| "learning_rate": 0.0001496639851775337, | |
| "loss": 1.6935, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.0907932150210805, | |
| "grad_norm": 8.133615493774414, | |
| "learning_rate": 0.00014946201788985358, | |
| "loss": 1.3022, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.0909893126777135, | |
| "grad_norm": 23.21516990661621, | |
| "learning_rate": 0.0001492597831410393, | |
| "loss": 1.9148, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.0911854103343465, | |
| "grad_norm": 5.873632431030273, | |
| "learning_rate": 0.00014905728202465595, | |
| "loss": 0.8053, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.09138150799097951, | |
| "grad_norm": 9.700281143188477, | |
| "learning_rate": 0.000148854515635709, | |
| "loss": 1.8187, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.09157760564761251, | |
| "grad_norm": 9.361727714538574, | |
| "learning_rate": 0.00014865148507063833, | |
| "loss": 1.5871, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.09177370330424552, | |
| "grad_norm": 6.021685600280762, | |
| "learning_rate": 0.00014844819142731223, | |
| "loss": 0.6632, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.09196980096087852, | |
| "grad_norm": 8.407185554504395, | |
| "learning_rate": 0.0001482446358050217, | |
| "loss": 2.1806, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.09216589861751152, | |
| "grad_norm": 6.5181450843811035, | |
| "learning_rate": 0.00014804081930447433, | |
| "loss": 1.1511, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.09236199627414453, | |
| "grad_norm": 6.047727584838867, | |
| "learning_rate": 0.00014783674302778832, | |
| "loss": 1.4668, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.09255809393077753, | |
| "grad_norm": 5.636353969573975, | |
| "learning_rate": 0.00014763240807848666, | |
| "loss": 1.2796, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.09275419158741054, | |
| "grad_norm": 10.02710247039795, | |
| "learning_rate": 0.00014742781556149103, | |
| "loss": 1.3822, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.09295028924404353, | |
| "grad_norm": 5.736359596252441, | |
| "learning_rate": 0.00014722296658311595, | |
| "loss": 1.6941, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.09314638690067653, | |
| "grad_norm": 4.80971622467041, | |
| "learning_rate": 0.0001470178622510627, | |
| "loss": 2.3505, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.09334248455730954, | |
| "grad_norm": 6.357244491577148, | |
| "learning_rate": 0.00014681250367441328, | |
| "loss": 3.0606, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.09353858221394254, | |
| "grad_norm": 6.115423202514648, | |
| "learning_rate": 0.00014660689196362462, | |
| "loss": 1.4211, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.09373467987057554, | |
| "grad_norm": 10.913143157958984, | |
| "learning_rate": 0.0001464010282305224, | |
| "loss": 1.6531, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.09393077752720855, | |
| "grad_norm": 7.621870517730713, | |
| "learning_rate": 0.000146194913588295, | |
| "loss": 1.964, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.09412687518384155, | |
| "grad_norm": 11.227121353149414, | |
| "learning_rate": 0.0001459885491514878, | |
| "loss": 1.639, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.09432297284047456, | |
| "grad_norm": 4.943283557891846, | |
| "learning_rate": 0.00014578193603599662, | |
| "loss": 0.5043, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.09451907049710756, | |
| "grad_norm": 13.006143569946289, | |
| "learning_rate": 0.0001455750753590622, | |
| "loss": 3.0382, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.09471516815374056, | |
| "grad_norm": 6.181257247924805, | |
| "learning_rate": 0.00014536796823926386, | |
| "loss": 1.3281, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.09491126581037357, | |
| "grad_norm": 4.05950403213501, | |
| "learning_rate": 0.0001451606157965136, | |
| "loss": 1.5701, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.09510736346700657, | |
| "grad_norm": 7.532302379608154, | |
| "learning_rate": 0.0001449530191520499, | |
| "loss": 1.811, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.09530346112363958, | |
| "grad_norm": 7.403939247131348, | |
| "learning_rate": 0.00014474517942843175, | |
| "loss": 1.8269, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.09549955878027258, | |
| "grad_norm": 18.15042495727539, | |
| "learning_rate": 0.0001445370977495326, | |
| "loss": 2.4586, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.09569565643690559, | |
| "grad_norm": 13.836175918579102, | |
| "learning_rate": 0.00014432877524053427, | |
| "loss": 1.7828, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.09589175409353858, | |
| "grad_norm": 5.9050822257995605, | |
| "learning_rate": 0.0001441202130279208, | |
| "loss": 2.7068, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.09608785175017158, | |
| "grad_norm": 7.173941612243652, | |
| "learning_rate": 0.0001439114122394724, | |
| "loss": 2.7424, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.09628394940680458, | |
| "grad_norm": 17.888187408447266, | |
| "learning_rate": 0.0001437023740042594, | |
| "loss": 2.0253, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.09648004706343759, | |
| "grad_norm": 6.492856025695801, | |
| "learning_rate": 0.00014349309945263606, | |
| "loss": 2.1604, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.09667614472007059, | |
| "grad_norm": 11.6004638671875, | |
| "learning_rate": 0.00014328358971623455, | |
| "loss": 2.1492, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.0968722423767036, | |
| "grad_norm": 6.262972354888916, | |
| "learning_rate": 0.00014307384592795872, | |
| "loss": 2.3007, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.0970683400333366, | |
| "grad_norm": 6.768704891204834, | |
| "learning_rate": 0.00014286386922197805, | |
| "loss": 0.6546, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.0972644376899696, | |
| "grad_norm": 4.372698783874512, | |
| "learning_rate": 0.0001426536607337215, | |
| "loss": 2.3647, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.09746053534660261, | |
| "grad_norm": 12.042864799499512, | |
| "learning_rate": 0.00014244322159987145, | |
| "loss": 2.1466, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.09765663300323561, | |
| "grad_norm": 5.374109745025635, | |
| "learning_rate": 0.0001422325529583573, | |
| "loss": 0.8599, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.09785273065986862, | |
| "grad_norm": 7.459904193878174, | |
| "learning_rate": 0.00014202165594834963, | |
| "loss": 1.7119, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.09804882831650162, | |
| "grad_norm": 8.25338363647461, | |
| "learning_rate": 0.00014181053171025392, | |
| "loss": 1.3014, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09824492597313463, | |
| "grad_norm": 8.594246864318848, | |
| "learning_rate": 0.00014159918138570424, | |
| "loss": 1.9332, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.09844102362976763, | |
| "grad_norm": 4.802338600158691, | |
| "learning_rate": 0.00014138760611755727, | |
| "loss": 1.6591, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.09863712128640063, | |
| "grad_norm": 5.049933433532715, | |
| "learning_rate": 0.00014117580704988612, | |
| "loss": 1.3722, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.09883321894303362, | |
| "grad_norm": 9.438610076904297, | |
| "learning_rate": 0.00014096378532797393, | |
| "loss": 2.1888, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.09902931659966663, | |
| "grad_norm": 11.144261360168457, | |
| "learning_rate": 0.00014075154209830792, | |
| "loss": 1.2526, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.09922541425629963, | |
| "grad_norm": 12.456216812133789, | |
| "learning_rate": 0.0001405390785085731, | |
| "loss": 2.8251, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.09942151191293264, | |
| "grad_norm": 13.78912353515625, | |
| "learning_rate": 0.00014032639570764593, | |
| "loss": 2.3312, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.09961760956956564, | |
| "grad_norm": 22.588031768798828, | |
| "learning_rate": 0.00014011349484558847, | |
| "loss": 1.8865, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.09981370722619864, | |
| "grad_norm": 5.357258319854736, | |
| "learning_rate": 0.00013990037707364166, | |
| "loss": 2.1296, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.10000980488283165, | |
| "grad_norm": 13.293752670288086, | |
| "learning_rate": 0.00013968704354421952, | |
| "loss": 2.0479, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.10020590253946465, | |
| "grad_norm": 6.764191150665283, | |
| "learning_rate": 0.00013947349541090274, | |
| "loss": 1.3908, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.10040200019609766, | |
| "grad_norm": 5.701320648193359, | |
| "learning_rate": 0.00013925973382843246, | |
| "loss": 1.1343, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.10059809785273066, | |
| "grad_norm": 6.728299140930176, | |
| "learning_rate": 0.0001390457599527039, | |
| "loss": 1.3219, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.10079419550936367, | |
| "grad_norm": 9.094908714294434, | |
| "learning_rate": 0.00013883157494076046, | |
| "loss": 1.1282, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.10099029316599667, | |
| "grad_norm": 7.321430683135986, | |
| "learning_rate": 0.00013861717995078708, | |
| "loss": 1.0628, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.10118639082262967, | |
| "grad_norm": 5.976141452789307, | |
| "learning_rate": 0.00013840257614210414, | |
| "loss": 1.5372, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.10138248847926268, | |
| "grad_norm": 7.5356926918029785, | |
| "learning_rate": 0.00013818776467516125, | |
| "loss": 2.0415, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.10157858613589568, | |
| "grad_norm": 5.656033992767334, | |
| "learning_rate": 0.00013797274671153092, | |
| "loss": 1.9171, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.10177468379252867, | |
| "grad_norm": 4.822179317474365, | |
| "learning_rate": 0.0001377575234139022, | |
| "loss": 1.1379, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.10197078144916168, | |
| "grad_norm": 2.5650954246520996, | |
| "learning_rate": 0.0001375420959460745, | |
| "loss": 0.3984, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.10216687910579468, | |
| "grad_norm": 6.997866630554199, | |
| "learning_rate": 0.00013732646547295126, | |
| "loss": 1.2149, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.10236297676242768, | |
| "grad_norm": 9.925884246826172, | |
| "learning_rate": 0.00013711063316053368, | |
| "loss": 1.9542, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.10255907441906069, | |
| "grad_norm": 7.913006782531738, | |
| "learning_rate": 0.00013689460017591432, | |
| "loss": 1.3162, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.1027551720756937, | |
| "grad_norm": 8.868382453918457, | |
| "learning_rate": 0.00013667836768727092, | |
| "loss": 2.8749, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.1029512697323267, | |
| "grad_norm": 7.418759346008301, | |
| "learning_rate": 0.00013646193686386, | |
| "loss": 1.2697, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.1031473673889597, | |
| "grad_norm": 9.501752853393555, | |
| "learning_rate": 0.00013624530887601055, | |
| "loss": 1.1473, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.1033434650455927, | |
| "grad_norm": 6.225274562835693, | |
| "learning_rate": 0.0001360284848951177, | |
| "loss": 1.0762, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.10353956270222571, | |
| "grad_norm": 6.502389907836914, | |
| "learning_rate": 0.0001358114660936364, | |
| "loss": 1.383, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.10373566035885871, | |
| "grad_norm": 17.53512954711914, | |
| "learning_rate": 0.00013559425364507508, | |
| "loss": 1.7968, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.10393175801549172, | |
| "grad_norm": 6.0248122215271, | |
| "learning_rate": 0.00013537684872398927, | |
| "loss": 1.2982, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.10412785567212472, | |
| "grad_norm": 4.621517658233643, | |
| "learning_rate": 0.00013515925250597537, | |
| "loss": 1.5811, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.10432395332875773, | |
| "grad_norm": 6.690089702606201, | |
| "learning_rate": 0.00013494146616766406, | |
| "loss": 1.3376, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.10452005098539073, | |
| "grad_norm": 6.418732643127441, | |
| "learning_rate": 0.00013472349088671418, | |
| "loss": 1.507, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.10471614864202373, | |
| "grad_norm": 12.271800994873047, | |
| "learning_rate": 0.0001345053278418062, | |
| "loss": 1.6329, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.10491224629865673, | |
| "grad_norm": 10.394608497619629, | |
| "learning_rate": 0.000134286978212636, | |
| "loss": 1.8793, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.10510834395528973, | |
| "grad_norm": 5.690524101257324, | |
| "learning_rate": 0.00013406844317990826, | |
| "loss": 2.2874, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.10530444161192273, | |
| "grad_norm": 4.313036918640137, | |
| "learning_rate": 0.00013384972392533034, | |
| "loss": 2.2905, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.10550053926855574, | |
| "grad_norm": 6.115211009979248, | |
| "learning_rate": 0.0001336308216316056, | |
| "loss": 2.3146, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.10569663692518874, | |
| "grad_norm": 7.43934965133667, | |
| "learning_rate": 0.0001334117374824274, | |
| "loss": 2.5837, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.10589273458182175, | |
| "grad_norm": 8.397297859191895, | |
| "learning_rate": 0.00013319247266247225, | |
| "loss": 2.1631, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.10608883223845475, | |
| "grad_norm": 5.273848056793213, | |
| "learning_rate": 0.00013297302835739376, | |
| "loss": 1.261, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.10628492989508775, | |
| "grad_norm": 5.789434909820557, | |
| "learning_rate": 0.00013275340575381598, | |
| "loss": 1.8465, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.10648102755172076, | |
| "grad_norm": 8.769668579101562, | |
| "learning_rate": 0.0001325336060393272, | |
| "loss": 1.3243, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.10667712520835376, | |
| "grad_norm": 7.573061943054199, | |
| "learning_rate": 0.00013231363040247334, | |
| "loss": 1.243, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.10687322286498677, | |
| "grad_norm": 9.284170150756836, | |
| "learning_rate": 0.00013209348003275165, | |
| "loss": 1.6408, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.10706932052161977, | |
| "grad_norm": 4.322136878967285, | |
| "learning_rate": 0.00013187315612060415, | |
| "loss": 3.2532, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.10726541817825277, | |
| "grad_norm": 7.865033149719238, | |
| "learning_rate": 0.00013165265985741142, | |
| "loss": 3.5798, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.10746151583488578, | |
| "grad_norm": 3.1747426986694336, | |
| "learning_rate": 0.00013143199243548587, | |
| "loss": 1.1817, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.10765761349151878, | |
| "grad_norm": 6.890042781829834, | |
| "learning_rate": 0.00013121115504806553, | |
| "loss": 2.0318, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.10785371114815177, | |
| "grad_norm": 7.632523059844971, | |
| "learning_rate": 0.00013099014888930748, | |
| "loss": 1.3925, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.10804980880478478, | |
| "grad_norm": 4.939361095428467, | |
| "learning_rate": 0.00013076897515428132, | |
| "loss": 1.3661, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.10824590646141778, | |
| "grad_norm": 6.433002948760986, | |
| "learning_rate": 0.00013054763503896294, | |
| "loss": 0.9917, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.10844200411805079, | |
| "grad_norm": 4.507363319396973, | |
| "learning_rate": 0.00013032612974022784, | |
| "loss": 0.9992, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.10863810177468379, | |
| "grad_norm": 9.477548599243164, | |
| "learning_rate": 0.00013010446045584479, | |
| "loss": 1.1149, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.1088341994313168, | |
| "grad_norm": 5.909473419189453, | |
| "learning_rate": 0.00012988262838446922, | |
| "loss": 2.0005, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.1090302970879498, | |
| "grad_norm": 8.154961585998535, | |
| "learning_rate": 0.00012966063472563685, | |
| "loss": 2.1173, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.1092263947445828, | |
| "grad_norm": 4.8719868659973145, | |
| "learning_rate": 0.00012943848067975718, | |
| "loss": 2.925, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.1094224924012158, | |
| "grad_norm": 5.895204544067383, | |
| "learning_rate": 0.00012921616744810698, | |
| "loss": 2.0375, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.10961859005784881, | |
| "grad_norm": 5.050942897796631, | |
| "learning_rate": 0.0001289936962328238, | |
| "loss": 1.6125, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.10981468771448182, | |
| "grad_norm": 5.890807151794434, | |
| "learning_rate": 0.0001287710682368995, | |
| "loss": 0.991, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.11001078537111482, | |
| "grad_norm": 5.316340446472168, | |
| "learning_rate": 0.00012854828466417364, | |
| "loss": 1.8846, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.11020688302774782, | |
| "grad_norm": 14.81187629699707, | |
| "learning_rate": 0.00012832534671932715, | |
| "loss": 2.141, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.11040298068438083, | |
| "grad_norm": 11.379783630371094, | |
| "learning_rate": 0.0001281022556078756, | |
| "loss": 2.9432, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.11059907834101383, | |
| "grad_norm": 17.28523826599121, | |
| "learning_rate": 0.0001278790125361629, | |
| "loss": 2.4028, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.11079517599764682, | |
| "grad_norm": 6.626383304595947, | |
| "learning_rate": 0.0001276556187113546, | |
| "loss": 1.1534, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.11099127365427983, | |
| "grad_norm": 7.6215128898620605, | |
| "learning_rate": 0.00012743207534143144, | |
| "loss": 1.6464, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.11118737131091283, | |
| "grad_norm": 8.348710060119629, | |
| "learning_rate": 0.00012720838363518286, | |
| "loss": 1.1823, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.11138346896754583, | |
| "grad_norm": 5.307929039001465, | |
| "learning_rate": 0.00012698454480220039, | |
| "loss": 2.3759, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.11157956662417884, | |
| "grad_norm": 4.690614223480225, | |
| "learning_rate": 0.00012676056005287106, | |
| "loss": 1.3698, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.11177566428081184, | |
| "grad_norm": 5.7534871101379395, | |
| "learning_rate": 0.00012653643059837107, | |
| "loss": 1.9587, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.11197176193744485, | |
| "grad_norm": 5.088260650634766, | |
| "learning_rate": 0.00012631215765065897, | |
| "loss": 1.0195, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.11216785959407785, | |
| "grad_norm": 5.987957000732422, | |
| "learning_rate": 0.0001260877424224693, | |
| "loss": 2.3612, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.11236395725071086, | |
| "grad_norm": 5.895928859710693, | |
| "learning_rate": 0.000125863186127306, | |
| "loss": 2.0393, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.11256005490734386, | |
| "grad_norm": 6.654543876647949, | |
| "learning_rate": 0.00012563848997943568, | |
| "loss": 1.7165, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.11275615256397686, | |
| "grad_norm": 13.193924903869629, | |
| "learning_rate": 0.00012541365519388138, | |
| "loss": 1.3969, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.11295225022060987, | |
| "grad_norm": 5.471423149108887, | |
| "learning_rate": 0.00012518868298641564, | |
| "loss": 1.091, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.11314834787724287, | |
| "grad_norm": 9.188228607177734, | |
| "learning_rate": 0.00012496357457355422, | |
| "loss": 2.0812, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.11334444553387588, | |
| "grad_norm": 11.622628211975098, | |
| "learning_rate": 0.00012473833117254931, | |
| "loss": 2.1117, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.11354054319050888, | |
| "grad_norm": 4.972383975982666, | |
| "learning_rate": 0.00012451295400138314, | |
| "loss": 1.3126, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.11373664084714187, | |
| "grad_norm": 7.87300968170166, | |
| "learning_rate": 0.0001242874442787611, | |
| "loss": 3.2112, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.11393273850377487, | |
| "grad_norm": 5.145363807678223, | |
| "learning_rate": 0.0001240618032241056, | |
| "loss": 1.2819, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.11412883616040788, | |
| "grad_norm": 12.539934158325195, | |
| "learning_rate": 0.00012383603205754894, | |
| "loss": 2.2262, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.11432493381704088, | |
| "grad_norm": 5.802661895751953, | |
| "learning_rate": 0.00012361013199992726, | |
| "loss": 1.9663, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.11452103147367389, | |
| "grad_norm": 16.773561477661133, | |
| "learning_rate": 0.0001233841042727734, | |
| "loss": 2.342, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.11471712913030689, | |
| "grad_norm": 6.163604259490967, | |
| "learning_rate": 0.0001231579500983108, | |
| "loss": 1.3009, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.1149132267869399, | |
| "grad_norm": 7.877258777618408, | |
| "learning_rate": 0.00012293167069944645, | |
| "loss": 1.9388, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.1151093244435729, | |
| "grad_norm": 13.269306182861328, | |
| "learning_rate": 0.00012270526729976465, | |
| "loss": 2.3028, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.1153054221002059, | |
| "grad_norm": 9.610801696777344, | |
| "learning_rate": 0.00012247874112352009, | |
| "loss": 1.0249, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.11550151975683891, | |
| "grad_norm": 11.20408821105957, | |
| "learning_rate": 0.00012225209339563145, | |
| "loss": 2.2807, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.11569761741347191, | |
| "grad_norm": 13.390026092529297, | |
| "learning_rate": 0.00012202532534167463, | |
| "loss": 1.7141, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.11589371507010492, | |
| "grad_norm": 4.4530229568481445, | |
| "learning_rate": 0.00012179843818787624, | |
| "loss": 1.4108, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.11608981272673792, | |
| "grad_norm": 12.671486854553223, | |
| "learning_rate": 0.00012157143316110684, | |
| "loss": 1.6154, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.11628591038337092, | |
| "grad_norm": 6.195436477661133, | |
| "learning_rate": 0.00012134431148887442, | |
| "loss": 3.4713, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.11648200804000393, | |
| "grad_norm": 5.860586643218994, | |
| "learning_rate": 0.00012111707439931776, | |
| "loss": 2.0938, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.11667810569663692, | |
| "grad_norm": 7.4623260498046875, | |
| "learning_rate": 0.00012088972312119964, | |
| "loss": 1.9674, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.11687420335326992, | |
| "grad_norm": 7.368410110473633, | |
| "learning_rate": 0.00012066225888390042, | |
| "loss": 1.3425, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.11707030100990293, | |
| "grad_norm": 7.527665138244629, | |
| "learning_rate": 0.00012043468291741116, | |
| "loss": 1.6366, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.11726639866653593, | |
| "grad_norm": 6.0890793800354, | |
| "learning_rate": 0.00012020699645232721, | |
| "loss": 1.9652, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.11746249632316894, | |
| "grad_norm": 6.350461959838867, | |
| "learning_rate": 0.00011997920071984133, | |
| "loss": 2.2867, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.11765859397980194, | |
| "grad_norm": 6.444486618041992, | |
| "learning_rate": 0.00011975129695173719, | |
| "loss": 0.7242, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.11785469163643494, | |
| "grad_norm": 7.270365238189697, | |
| "learning_rate": 0.00011952328638038261, | |
| "loss": 2.2346, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.11805078929306795, | |
| "grad_norm": 5.961056709289551, | |
| "learning_rate": 0.00011929517023872298, | |
| "loss": 2.4926, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.11824688694970095, | |
| "grad_norm": 5.347055435180664, | |
| "learning_rate": 0.00011906694976027455, | |
| "loss": 1.3998, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.11844298460633396, | |
| "grad_norm": 11.233309745788574, | |
| "learning_rate": 0.0001188386261791177, | |
| "loss": 2.0185, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.11863908226296696, | |
| "grad_norm": 4.813647747039795, | |
| "learning_rate": 0.0001186102007298904, | |
| "loss": 1.3932, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.11883517991959996, | |
| "grad_norm": 5.281280040740967, | |
| "learning_rate": 0.00011838167464778142, | |
| "loss": 1.7445, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.11903127757623297, | |
| "grad_norm": 8.755961418151855, | |
| "learning_rate": 0.00011815304916852372, | |
| "loss": 1.656, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.11922737523286597, | |
| "grad_norm": 15.501289367675781, | |
| "learning_rate": 0.00011792432552838772, | |
| "loss": 2.0894, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.11942347288949898, | |
| "grad_norm": 6.1117753982543945, | |
| "learning_rate": 0.00011769550496417466, | |
| "loss": 1.0473, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.11961957054613197, | |
| "grad_norm": 9.669829368591309, | |
| "learning_rate": 0.00011746658871320983, | |
| "loss": 0.7436, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.11981566820276497, | |
| "grad_norm": 10.253652572631836, | |
| "learning_rate": 0.00011723757801333601, | |
| "loss": 2.7502, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.12001176585939798, | |
| "grad_norm": 7.084132671356201, | |
| "learning_rate": 0.00011700847410290667, | |
| "loss": 2.5537, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.12020786351603098, | |
| "grad_norm": 5.936439037322998, | |
| "learning_rate": 0.0001167792782207793, | |
| "loss": 1.1551, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.12040396117266398, | |
| "grad_norm": 7.053248405456543, | |
| "learning_rate": 0.0001165499916063087, | |
| "loss": 1.2488, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.12060005882929699, | |
| "grad_norm": 7.141607284545898, | |
| "learning_rate": 0.00011632061549934036, | |
| "loss": 1.4092, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.12079615648592999, | |
| "grad_norm": 7.723854064941406, | |
| "learning_rate": 0.00011609115114020363, | |
| "loss": 1.7669, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.120992254142563, | |
| "grad_norm": 7.095085144042969, | |
| "learning_rate": 0.00011586159976970514, | |
| "loss": 1.5018, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.121188351799196, | |
| "grad_norm": 6.469241142272949, | |
| "learning_rate": 0.00011563196262912193, | |
| "loss": 2.3377, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.121384449455829, | |
| "grad_norm": 6.16892671585083, | |
| "learning_rate": 0.00011540224096019494, | |
| "loss": 2.0011, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.12158054711246201, | |
| "grad_norm": 4.764594554901123, | |
| "learning_rate": 0.00011517243600512209, | |
| "loss": 0.8995, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.12177664476909501, | |
| "grad_norm": 8.256946563720703, | |
| "learning_rate": 0.00011494254900655175, | |
| "loss": 1.5867, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.12197274242572802, | |
| "grad_norm": 9.220061302185059, | |
| "learning_rate": 0.00011471258120757586, | |
| "loss": 1.0146, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.12216884008236102, | |
| "grad_norm": 6.284097194671631, | |
| "learning_rate": 0.00011448253385172335, | |
| "loss": 0.8305, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.12236493773899403, | |
| "grad_norm": 6.183596134185791, | |
| "learning_rate": 0.00011425240818295326, | |
| "loss": 0.7749, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.12256103539562702, | |
| "grad_norm": 7.550481796264648, | |
| "learning_rate": 0.00011402220544564819, | |
| "loss": 1.3157, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.12275713305226002, | |
| "grad_norm": 4.810642719268799, | |
| "learning_rate": 0.0001137919268846074, | |
| "loss": 1.6442, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.12295323070889302, | |
| "grad_norm": 10.603952407836914, | |
| "learning_rate": 0.00011356157374504027, | |
| "loss": 1.7275, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.12314932836552603, | |
| "grad_norm": 13.009543418884277, | |
| "learning_rate": 0.00011333114727255929, | |
| "loss": 2.3718, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.12334542602215903, | |
| "grad_norm": 8.453059196472168, | |
| "learning_rate": 0.00011310064871317366, | |
| "loss": 1.6558, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.12354152367879204, | |
| "grad_norm": 7.005093574523926, | |
| "learning_rate": 0.00011287007931328226, | |
| "loss": 1.9709, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.12373762133542504, | |
| "grad_norm": 5.710804462432861, | |
| "learning_rate": 0.00011263944031966714, | |
| "loss": 1.8595, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.12393371899205805, | |
| "grad_norm": 3.608022689819336, | |
| "learning_rate": 0.00011240873297948653, | |
| "loss": 1.7165, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.12412981664869105, | |
| "grad_norm": 8.973870277404785, | |
| "learning_rate": 0.0001121779585402684, | |
| "loss": 1.4924, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.12432591430532405, | |
| "grad_norm": 4.735976219177246, | |
| "learning_rate": 0.00011194711824990344, | |
| "loss": 1.2721, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.12452201196195706, | |
| "grad_norm": 4.169306755065918, | |
| "learning_rate": 0.00011171621335663844, | |
| "loss": 0.718, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.12471810961859006, | |
| "grad_norm": 14.129206657409668, | |
| "learning_rate": 0.00011148524510906956, | |
| "loss": 1.1904, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.12491420727522307, | |
| "grad_norm": 12.76352310180664, | |
| "learning_rate": 0.00011125421475613554, | |
| "loss": 2.4872, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.12511030493185607, | |
| "grad_norm": 8.636000633239746, | |
| "learning_rate": 0.00011102312354711091, | |
| "loss": 0.7629, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.12530640258848907, | |
| "grad_norm": 6.22203254699707, | |
| "learning_rate": 0.00011079197273159925, | |
| "loss": 1.7372, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.12550250024512208, | |
| "grad_norm": 4.235958576202393, | |
| "learning_rate": 0.0001105607635595266, | |
| "loss": 1.8661, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.12569859790175508, | |
| "grad_norm": 5.598705768585205, | |
| "learning_rate": 0.00011032949728113437, | |
| "loss": 1.7918, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.1258946955583881, | |
| "grad_norm": 9.218809127807617, | |
| "learning_rate": 0.00011009817514697291, | |
| "loss": 1.2968, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.1260907932150211, | |
| "grad_norm": 7.068932056427002, | |
| "learning_rate": 0.00010986679840789451, | |
| "loss": 1.8485, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.1262868908716541, | |
| "grad_norm": 3.891483783721924, | |
| "learning_rate": 0.00010963536831504684, | |
| "loss": 0.3918, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.1264829885282871, | |
| "grad_norm": 5.959576606750488, | |
| "learning_rate": 0.00010940388611986592, | |
| "loss": 0.7057, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.1266790861849201, | |
| "grad_norm": 11.663713455200195, | |
| "learning_rate": 0.00010917235307406966, | |
| "loss": 1.5916, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.12687518384155308, | |
| "grad_norm": 15.05775260925293, | |
| "learning_rate": 0.00010894077042965083, | |
| "loss": 2.979, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.12707128149818608, | |
| "grad_norm": 7.2112650871276855, | |
| "learning_rate": 0.00010870913943887049, | |
| "loss": 1.6293, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.1272673791548191, | |
| "grad_norm": 13.646367073059082, | |
| "learning_rate": 0.00010847746135425102, | |
| "loss": 2.5277, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.1274634768114521, | |
| "grad_norm": 6.29921293258667, | |
| "learning_rate": 0.0001082457374285696, | |
| "loss": 2.4015, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.1276595744680851, | |
| "grad_norm": 6.337920188903809, | |
| "learning_rate": 0.00010801396891485114, | |
| "loss": 0.9928, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.1278556721247181, | |
| "grad_norm": 7.32103157043457, | |
| "learning_rate": 0.00010778215706636177, | |
| "loss": 2.4466, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.1280517697813511, | |
| "grad_norm": 10.154195785522461, | |
| "learning_rate": 0.00010755030313660188, | |
| "loss": 2.4858, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.1282478674379841, | |
| "grad_norm": 7.129855632781982, | |
| "learning_rate": 0.00010731840837929946, | |
| "loss": 1.2546, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.1284439650946171, | |
| "grad_norm": 5.403641223907471, | |
| "learning_rate": 0.00010708647404840319, | |
| "loss": 1.1017, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.12864006275125012, | |
| "grad_norm": 5.554705619812012, | |
| "learning_rate": 0.00010685450139807584, | |
| "loss": 2.0789, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.12883616040788312, | |
| "grad_norm": 7.935024738311768, | |
| "learning_rate": 0.00010662249168268736, | |
| "loss": 1.9944, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.12903225806451613, | |
| "grad_norm": 4.453607559204102, | |
| "learning_rate": 0.00010639044615680809, | |
| "loss": 0.8442, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.12922835572114913, | |
| "grad_norm": 13.252726554870605, | |
| "learning_rate": 0.00010615836607520202, | |
| "loss": 1.1374, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.12942445337778213, | |
| "grad_norm": 4.426390171051025, | |
| "learning_rate": 0.00010592625269282, | |
| "loss": 1.0554, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.12962055103441514, | |
| "grad_norm": 6.692317008972168, | |
| "learning_rate": 0.000105694107264793, | |
| "loss": 0.6669, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.12981664869104814, | |
| "grad_norm": 6.067706108093262, | |
| "learning_rate": 0.00010546193104642519, | |
| "loss": 1.488, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.13001274634768115, | |
| "grad_norm": 7.749240875244141, | |
| "learning_rate": 0.00010522972529318733, | |
| "loss": 1.8632, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.13020884400431415, | |
| "grad_norm": 4.964468955993652, | |
| "learning_rate": 0.00010499749126070979, | |
| "loss": 0.5845, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.13040494166094715, | |
| "grad_norm": 6.432995319366455, | |
| "learning_rate": 0.00010476523020477593, | |
| "loss": 1.1617, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.13060103931758016, | |
| "grad_norm": 6.4099249839782715, | |
| "learning_rate": 0.00010453294338131519, | |
| "loss": 1.2821, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.13079713697421316, | |
| "grad_norm": 5.935438632965088, | |
| "learning_rate": 0.0001043006320463964, | |
| "loss": 1.4453, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.13099323463084617, | |
| "grad_norm": 10.963349342346191, | |
| "learning_rate": 0.00010406829745622085, | |
| "loss": 1.6318, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.13118933228747917, | |
| "grad_norm": 9.157288551330566, | |
| "learning_rate": 0.00010383594086711567, | |
| "loss": 1.6981, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.13138542994411218, | |
| "grad_norm": 17.15442657470703, | |
| "learning_rate": 0.00010360356353552687, | |
| "loss": 1.498, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.13158152760074518, | |
| "grad_norm": 3.763289451599121, | |
| "learning_rate": 0.00010337116671801272, | |
| "loss": 2.425, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.13177762525737818, | |
| "grad_norm": 8.91812801361084, | |
| "learning_rate": 0.00010313875167123672, | |
| "loss": 2.3129, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.1319737229140112, | |
| "grad_norm": 23.093202590942383, | |
| "learning_rate": 0.00010290631965196109, | |
| "loss": 2.3172, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.1321698205706442, | |
| "grad_norm": 6.487079620361328, | |
| "learning_rate": 0.00010267387191703972, | |
| "loss": 1.6108, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.1323659182272772, | |
| "grad_norm": 3.9093501567840576, | |
| "learning_rate": 0.00010244140972341155, | |
| "loss": 0.8954, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.1325620158839102, | |
| "grad_norm": 11.078289985656738, | |
| "learning_rate": 0.00010220893432809365, | |
| "loss": 2.0604, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.13275811354054318, | |
| "grad_norm": 7.805664539337158, | |
| "learning_rate": 0.00010197644698817446, | |
| "loss": 1.834, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.13295421119717618, | |
| "grad_norm": 6.838125705718994, | |
| "learning_rate": 0.00010174394896080713, | |
| "loss": 1.4876, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.13315030885380919, | |
| "grad_norm": 3.9814906120300293, | |
| "learning_rate": 0.00010151144150320247, | |
| "loss": 1.3479, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.1333464065104422, | |
| "grad_norm": 6.396786212921143, | |
| "learning_rate": 0.00010127892587262233, | |
| "loss": 1.7642, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.1335425041670752, | |
| "grad_norm": 10.133893013000488, | |
| "learning_rate": 0.00010104640332637278, | |
| "loss": 0.8957, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.1337386018237082, | |
| "grad_norm": 6.3815412521362305, | |
| "learning_rate": 0.00010081387512179729, | |
| "loss": 1.4104, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.1337386018237082, | |
| "eval_loss": 0.41039586067199707, | |
| "eval_runtime": 77.8442, | |
| "eval_samples_per_second": 27.594, | |
| "eval_steps_per_second": 13.797, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.1339346994803412, | |
| "grad_norm": 11.698119163513184, | |
| "learning_rate": 0.00010058134251626983, | |
| "loss": 2.1025, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.1341307971369742, | |
| "grad_norm": 5.185816287994385, | |
| "learning_rate": 0.00010034880676718832, | |
| "loss": 1.0986, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.1343268947936072, | |
| "grad_norm": 6.936627388000488, | |
| "learning_rate": 0.0001001162691319675, | |
| "loss": 1.3077, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.13452299245024021, | |
| "grad_norm": 4.661152362823486, | |
| "learning_rate": 9.988373086803251e-05, | |
| "loss": 1.6312, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.13471909010687322, | |
| "grad_norm": 4.227341651916504, | |
| "learning_rate": 9.965119323281174e-05, | |
| "loss": 0.6856, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.13491518776350622, | |
| "grad_norm": 24.00637435913086, | |
| "learning_rate": 9.941865748373017e-05, | |
| "loss": 2.8267, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.13511128542013923, | |
| "grad_norm": 4.701898574829102, | |
| "learning_rate": 9.918612487820273e-05, | |
| "loss": 1.7314, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.13530738307677223, | |
| "grad_norm": 7.700753688812256, | |
| "learning_rate": 9.895359667362724e-05, | |
| "loss": 1.1831, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.13550348073340523, | |
| "grad_norm": 5.045268535614014, | |
| "learning_rate": 9.87210741273777e-05, | |
| "loss": 2.6148, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.13569957839003824, | |
| "grad_norm": 8.903031349182129, | |
| "learning_rate": 9.848855849679754e-05, | |
| "loss": 2.3442, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.13589567604667124, | |
| "grad_norm": 6.869392395019531, | |
| "learning_rate": 9.82560510391929e-05, | |
| "loss": 1.7329, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.13609177370330425, | |
| "grad_norm": 5.181933403015137, | |
| "learning_rate": 9.802355301182556e-05, | |
| "loss": 2.2152, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.13628787135993725, | |
| "grad_norm": 9.427200317382812, | |
| "learning_rate": 9.779106567190642e-05, | |
| "loss": 0.9606, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.13648396901657026, | |
| "grad_norm": 5.893560409545898, | |
| "learning_rate": 9.755859027658848e-05, | |
| "loss": 0.7759, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.13668006667320326, | |
| "grad_norm": 3.864348888397217, | |
| "learning_rate": 9.73261280829603e-05, | |
| "loss": 0.714, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.13687616432983626, | |
| "grad_norm": 7.077169418334961, | |
| "learning_rate": 9.709368034803892e-05, | |
| "loss": 1.9489, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.13707226198646927, | |
| "grad_norm": 10.102705955505371, | |
| "learning_rate": 9.686124832876328e-05, | |
| "loss": 2.2884, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.13726835964310227, | |
| "grad_norm": 6.154596328735352, | |
| "learning_rate": 9.662883328198731e-05, | |
| "loss": 1.1614, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.13746445729973528, | |
| "grad_norm": 8.005276679992676, | |
| "learning_rate": 9.639643646447316e-05, | |
| "loss": 2.5097, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.13766055495636828, | |
| "grad_norm": 12.45338249206543, | |
| "learning_rate": 9.616405913288438e-05, | |
| "loss": 0.8686, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.13785665261300128, | |
| "grad_norm": 4.405421257019043, | |
| "learning_rate": 9.593170254377916e-05, | |
| "loss": 0.8499, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.1380527502696343, | |
| "grad_norm": 9.178966522216797, | |
| "learning_rate": 9.569936795360363e-05, | |
| "loss": 1.7378, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.1382488479262673, | |
| "grad_norm": 9.432720184326172, | |
| "learning_rate": 9.546705661868484e-05, | |
| "loss": 1.5712, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.1384449455829003, | |
| "grad_norm": 5.602529048919678, | |
| "learning_rate": 9.523476979522411e-05, | |
| "loss": 1.8486, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.1386410432395333, | |
| "grad_norm": 7.334858417510986, | |
| "learning_rate": 9.50025087392902e-05, | |
| "loss": 1.3093, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.13883714089616628, | |
| "grad_norm": 16.70349884033203, | |
| "learning_rate": 9.477027470681269e-05, | |
| "loss": 1.8117, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.13903323855279928, | |
| "grad_norm": 4.915411472320557, | |
| "learning_rate": 9.453806895357482e-05, | |
| "loss": 1.296, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.1392293362094323, | |
| "grad_norm": 9.790595054626465, | |
| "learning_rate": 9.430589273520703e-05, | |
| "loss": 1.7763, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.1394254338660653, | |
| "grad_norm": 5.754830837249756, | |
| "learning_rate": 9.407374730718e-05, | |
| "loss": 1.2865, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.1396215315226983, | |
| "grad_norm": 6.5165534019470215, | |
| "learning_rate": 9.3841633924798e-05, | |
| "loss": 1.0498, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.1398176291793313, | |
| "grad_norm": 4.821907043457031, | |
| "learning_rate": 9.360955384319195e-05, | |
| "loss": 1.1823, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.1400137268359643, | |
| "grad_norm": 7.226667881011963, | |
| "learning_rate": 9.337750831731266e-05, | |
| "loss": 1.0176, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.1402098244925973, | |
| "grad_norm": 5.560497760772705, | |
| "learning_rate": 9.314549860192417e-05, | |
| "loss": 1.1645, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.1404059221492303, | |
| "grad_norm": 4.834329128265381, | |
| "learning_rate": 9.291352595159682e-05, | |
| "loss": 1.682, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.14060201980586332, | |
| "grad_norm": 6.791970729827881, | |
| "learning_rate": 9.268159162070058e-05, | |
| "loss": 1.5408, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.14079811746249632, | |
| "grad_norm": 5.209107875823975, | |
| "learning_rate": 9.244969686339813e-05, | |
| "loss": 0.6371, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.14099421511912932, | |
| "grad_norm": 7.674073696136475, | |
| "learning_rate": 9.221784293363824e-05, | |
| "loss": 1.9635, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.14119031277576233, | |
| "grad_norm": 13.054336547851562, | |
| "learning_rate": 9.198603108514888e-05, | |
| "loss": 1.2977, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.14138641043239533, | |
| "grad_norm": 7.0277299880981445, | |
| "learning_rate": 9.175426257143045e-05, | |
| "loss": 1.3063, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.14158250808902834, | |
| "grad_norm": 10.085116386413574, | |
| "learning_rate": 9.152253864574898e-05, | |
| "loss": 1.5895, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.14177860574566134, | |
| "grad_norm": 11.588119506835938, | |
| "learning_rate": 9.129086056112955e-05, | |
| "loss": 1.6528, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.14197470340229434, | |
| "grad_norm": 6.269704818725586, | |
| "learning_rate": 9.10592295703492e-05, | |
| "loss": 1.1952, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.14217080105892735, | |
| "grad_norm": 5.680713176727295, | |
| "learning_rate": 9.082764692593039e-05, | |
| "loss": 2.322, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.14236689871556035, | |
| "grad_norm": 6.14839506149292, | |
| "learning_rate": 9.059611388013408e-05, | |
| "loss": 2.6696, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.14256299637219336, | |
| "grad_norm": 13.12667179107666, | |
| "learning_rate": 9.03646316849532e-05, | |
| "loss": 2.0335, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.14275909402882636, | |
| "grad_norm": 7.358575820922852, | |
| "learning_rate": 9.01332015921055e-05, | |
| "loss": 0.8538, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.14295519168545937, | |
| "grad_norm": 11.932861328125, | |
| "learning_rate": 8.990182485302713e-05, | |
| "loss": 1.4472, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.14315128934209237, | |
| "grad_norm": 8.689105987548828, | |
| "learning_rate": 8.967050271886564e-05, | |
| "loss": 1.8314, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.14334738699872537, | |
| "grad_norm": 5.939968585968018, | |
| "learning_rate": 8.943923644047342e-05, | |
| "loss": 0.8076, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.14354348465535838, | |
| "grad_norm": 8.181587219238281, | |
| "learning_rate": 8.920802726840076e-05, | |
| "loss": 1.4664, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.14373958231199138, | |
| "grad_norm": 7.449463844299316, | |
| "learning_rate": 8.897687645288914e-05, | |
| "loss": 1.0997, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.14393567996862439, | |
| "grad_norm": 5.9800848960876465, | |
| "learning_rate": 8.874578524386447e-05, | |
| "loss": 0.9288, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.1441317776252574, | |
| "grad_norm": 8.202332496643066, | |
| "learning_rate": 8.851475489093045e-05, | |
| "loss": 1.5658, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.1443278752818904, | |
| "grad_norm": 4.058600425720215, | |
| "learning_rate": 8.828378664336158e-05, | |
| "loss": 1.1746, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.1445239729385234, | |
| "grad_norm": 5.467014789581299, | |
| "learning_rate": 8.805288175009659e-05, | |
| "loss": 0.9683, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.14472007059515637, | |
| "grad_norm": 8.572188377380371, | |
| "learning_rate": 8.782204145973162e-05, | |
| "loss": 1.3502, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.14491616825178938, | |
| "grad_norm": 6.00791072845459, | |
| "learning_rate": 8.759126702051348e-05, | |
| "loss": 1.5011, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.14511226590842238, | |
| "grad_norm": 7.9531989097595215, | |
| "learning_rate": 8.736055968033289e-05, | |
| "loss": 2.3835, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.1453083635650554, | |
| "grad_norm": 7.390120029449463, | |
| "learning_rate": 8.712992068671774e-05, | |
| "loss": 1.1469, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.1455044612216884, | |
| "grad_norm": 6.503965377807617, | |
| "learning_rate": 8.689935128682635e-05, | |
| "loss": 1.9587, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.1457005588783214, | |
| "grad_norm": 7.921380996704102, | |
| "learning_rate": 8.666885272744073e-05, | |
| "loss": 1.9997, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.1458966565349544, | |
| "grad_norm": 7.979074001312256, | |
| "learning_rate": 8.643842625495978e-05, | |
| "loss": 1.4087, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.1460927541915874, | |
| "grad_norm": 5.653082847595215, | |
| "learning_rate": 8.620807311539259e-05, | |
| "loss": 1.3974, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.1462888518482204, | |
| "grad_norm": 7.764763355255127, | |
| "learning_rate": 8.597779455435183e-05, | |
| "loss": 1.4269, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.1464849495048534, | |
| "grad_norm": 11.947178840637207, | |
| "learning_rate": 8.574759181704676e-05, | |
| "loss": 2.4265, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.14668104716148642, | |
| "grad_norm": 9.1488037109375, | |
| "learning_rate": 8.551746614827669e-05, | |
| "loss": 2.2206, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.14687714481811942, | |
| "grad_norm": 5.008571147918701, | |
| "learning_rate": 8.528741879242415e-05, | |
| "loss": 2.1277, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.14707324247475242, | |
| "grad_norm": 4.040747165679932, | |
| "learning_rate": 8.505745099344827e-05, | |
| "loss": 0.7625, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.14726934013138543, | |
| "grad_norm": 8.826830863952637, | |
| "learning_rate": 8.482756399487793e-05, | |
| "loss": 1.1674, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.14746543778801843, | |
| "grad_norm": 4.931051731109619, | |
| "learning_rate": 8.45977590398051e-05, | |
| "loss": 1.1831, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.14766153544465144, | |
| "grad_norm": 5.232216835021973, | |
| "learning_rate": 8.436803737087806e-05, | |
| "loss": 1.0998, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.14785763310128444, | |
| "grad_norm": 9.30781364440918, | |
| "learning_rate": 8.413840023029488e-05, | |
| "loss": 2.8714, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.14805373075791745, | |
| "grad_norm": 6.524617671966553, | |
| "learning_rate": 8.390884885979639e-05, | |
| "loss": 1.3297, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.14824982841455045, | |
| "grad_norm": 6.3461432456970215, | |
| "learning_rate": 8.367938450065967e-05, | |
| "loss": 1.6513, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.14844592607118345, | |
| "grad_norm": 9.379528999328613, | |
| "learning_rate": 8.345000839369132e-05, | |
| "loss": 1.3968, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.14864202372781646, | |
| "grad_norm": 9.503636360168457, | |
| "learning_rate": 8.322072177922073e-05, | |
| "loss": 1.0089, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.14883812138444946, | |
| "grad_norm": 4.907662868499756, | |
| "learning_rate": 8.299152589709336e-05, | |
| "loss": 2.4323, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.14903421904108247, | |
| "grad_norm": 7.7067084312438965, | |
| "learning_rate": 8.2762421986664e-05, | |
| "loss": 1.4224, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.14923031669771547, | |
| "grad_norm": 5.226773738861084, | |
| "learning_rate": 8.253341128679018e-05, | |
| "loss": 2.0016, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.14942641435434847, | |
| "grad_norm": 8.012574195861816, | |
| "learning_rate": 8.230449503582538e-05, | |
| "loss": 1.8039, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.14962251201098148, | |
| "grad_norm": 6.705967426300049, | |
| "learning_rate": 8.207567447161231e-05, | |
| "loss": 2.5853, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.14981860966761448, | |
| "grad_norm": 9.187888145446777, | |
| "learning_rate": 8.184695083147627e-05, | |
| "loss": 1.0529, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.1500147073242475, | |
| "grad_norm": 4.542299270629883, | |
| "learning_rate": 8.16183253522186e-05, | |
| "loss": 1.2803, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.1502108049808805, | |
| "grad_norm": 5.094648361206055, | |
| "learning_rate": 8.138979927010964e-05, | |
| "loss": 0.9061, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.1504069026375135, | |
| "grad_norm": 13.718326568603516, | |
| "learning_rate": 8.116137382088233e-05, | |
| "loss": 2.0593, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.15060300029414647, | |
| "grad_norm": 6.143137454986572, | |
| "learning_rate": 8.093305023972546e-05, | |
| "loss": 0.8494, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.15079909795077948, | |
| "grad_norm": 6.3516154289245605, | |
| "learning_rate": 8.070482976127703e-05, | |
| "loss": 2.214, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.15099519560741248, | |
| "grad_norm": 9.909046173095703, | |
| "learning_rate": 8.047671361961741e-05, | |
| "loss": 0.928, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.15119129326404548, | |
| "grad_norm": 6.7013373374938965, | |
| "learning_rate": 8.024870304826286e-05, | |
| "loss": 1.2937, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.1513873909206785, | |
| "grad_norm": 7.552654266357422, | |
| "learning_rate": 8.002079928015867e-05, | |
| "loss": 2.3988, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.1515834885773115, | |
| "grad_norm": 4.806221961975098, | |
| "learning_rate": 7.97930035476728e-05, | |
| "loss": 1.9656, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.1517795862339445, | |
| "grad_norm": 13.320466041564941, | |
| "learning_rate": 7.956531708258887e-05, | |
| "loss": 2.1333, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.1519756838905775, | |
| "grad_norm": 7.802790641784668, | |
| "learning_rate": 7.933774111609964e-05, | |
| "loss": 0.766, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.1521717815472105, | |
| "grad_norm": 9.146486282348633, | |
| "learning_rate": 7.911027687880037e-05, | |
| "loss": 1.0215, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.1523678792038435, | |
| "grad_norm": 5.029911518096924, | |
| "learning_rate": 7.888292560068226e-05, | |
| "loss": 0.6491, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.1525639768604765, | |
| "grad_norm": 16.61541175842285, | |
| "learning_rate": 7.865568851112559e-05, | |
| "loss": 2.3993, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.15276007451710952, | |
| "grad_norm": 8.573732376098633, | |
| "learning_rate": 7.842856683889321e-05, | |
| "loss": 1.5773, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.15295617217374252, | |
| "grad_norm": 7.390541076660156, | |
| "learning_rate": 7.820156181212379e-05, | |
| "loss": 0.6461, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.15315226983037553, | |
| "grad_norm": 4.455470085144043, | |
| "learning_rate": 7.79746746583254e-05, | |
| "loss": 1.9557, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.15334836748700853, | |
| "grad_norm": 9.000947952270508, | |
| "learning_rate": 7.774790660436858e-05, | |
| "loss": 2.5368, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.15354446514364153, | |
| "grad_norm": 7.608001232147217, | |
| "learning_rate": 7.752125887647992e-05, | |
| "loss": 1.607, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.15374056280027454, | |
| "grad_norm": 4.594658851623535, | |
| "learning_rate": 7.729473270023537e-05, | |
| "loss": 0.9085, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.15393666045690754, | |
| "grad_norm": 4.750558853149414, | |
| "learning_rate": 7.706832930055357e-05, | |
| "loss": 1.8871, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.15413275811354055, | |
| "grad_norm": 6.5569987297058105, | |
| "learning_rate": 7.684204990168925e-05, | |
| "loss": 1.4386, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.15432885577017355, | |
| "grad_norm": 4.655035495758057, | |
| "learning_rate": 7.66158957272266e-05, | |
| "loss": 1.9621, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.15452495342680655, | |
| "grad_norm": 9.556004524230957, | |
| "learning_rate": 7.638986800007277e-05, | |
| "loss": 0.9825, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.15472105108343956, | |
| "grad_norm": 4.108066082000732, | |
| "learning_rate": 7.616396794245107e-05, | |
| "loss": 2.0709, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.15491714874007256, | |
| "grad_norm": 7.94503116607666, | |
| "learning_rate": 7.593819677589445e-05, | |
| "loss": 1.1699, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.15511324639670557, | |
| "grad_norm": 6.220666408538818, | |
| "learning_rate": 7.571255572123889e-05, | |
| "loss": 1.1849, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.15530934405333857, | |
| "grad_norm": 6.629292964935303, | |
| "learning_rate": 7.54870459986169e-05, | |
| "loss": 1.3573, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.15550544170997158, | |
| "grad_norm": 4.624551296234131, | |
| "learning_rate": 7.52616688274507e-05, | |
| "loss": 1.1043, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.15570153936660458, | |
| "grad_norm": 8.356186866760254, | |
| "learning_rate": 7.503642542644581e-05, | |
| "loss": 1.3448, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.15589763702323758, | |
| "grad_norm": 5.954531192779541, | |
| "learning_rate": 7.481131701358434e-05, | |
| "loss": 1.2289, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.1560937346798706, | |
| "grad_norm": 8.886476516723633, | |
| "learning_rate": 7.458634480611864e-05, | |
| "loss": 2.0398, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.1562898323365036, | |
| "grad_norm": 5.7918524742126465, | |
| "learning_rate": 7.436151002056433e-05, | |
| "loss": 1.4364, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.15648592999313657, | |
| "grad_norm": 9.434075355529785, | |
| "learning_rate": 7.413681387269406e-05, | |
| "loss": 1.8601, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.15668202764976957, | |
| "grad_norm": 5.067231178283691, | |
| "learning_rate": 7.391225757753068e-05, | |
| "loss": 1.414, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.15687812530640258, | |
| "grad_norm": 6.89833927154541, | |
| "learning_rate": 7.368784234934104e-05, | |
| "loss": 1.6815, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.15707422296303558, | |
| "grad_norm": 15.612439155578613, | |
| "learning_rate": 7.346356940162895e-05, | |
| "loss": 3.0339, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.15727032061966859, | |
| "grad_norm": 5.933378219604492, | |
| "learning_rate": 7.323943994712895e-05, | |
| "loss": 1.6531, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.1574664182763016, | |
| "grad_norm": 4.7962727546691895, | |
| "learning_rate": 7.301545519779964e-05, | |
| "loss": 1.8519, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.1576625159329346, | |
| "grad_norm": 7.490151882171631, | |
| "learning_rate": 7.279161636481715e-05, | |
| "loss": 1.429, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.1578586135895676, | |
| "grad_norm": 10.799139976501465, | |
| "learning_rate": 7.256792465856858e-05, | |
| "loss": 1.391, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.1580547112462006, | |
| "grad_norm": 4.731945514678955, | |
| "learning_rate": 7.23443812886454e-05, | |
| "loss": 1.3097, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.1582508089028336, | |
| "grad_norm": 6.669147491455078, | |
| "learning_rate": 7.21209874638371e-05, | |
| "loss": 2.5443, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.1584469065594666, | |
| "grad_norm": 7.046877861022949, | |
| "learning_rate": 7.189774439212442e-05, | |
| "loss": 1.5809, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.15864300421609961, | |
| "grad_norm": 5.848972320556641, | |
| "learning_rate": 7.167465328067289e-05, | |
| "loss": 1.6708, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.15883910187273262, | |
| "grad_norm": 6.136227607727051, | |
| "learning_rate": 7.145171533582636e-05, | |
| "loss": 1.2122, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.15903519952936562, | |
| "grad_norm": 4.7808709144592285, | |
| "learning_rate": 7.122893176310052e-05, | |
| "loss": 1.8965, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.15923129718599863, | |
| "grad_norm": 4.156777381896973, | |
| "learning_rate": 7.100630376717621e-05, | |
| "loss": 1.2742, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.15942739484263163, | |
| "grad_norm": 5.69835090637207, | |
| "learning_rate": 7.078383255189304e-05, | |
| "loss": 0.7664, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.15962349249926464, | |
| "grad_norm": 9.707277297973633, | |
| "learning_rate": 7.056151932024282e-05, | |
| "loss": 1.3492, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.15981959015589764, | |
| "grad_norm": 10.071001052856445, | |
| "learning_rate": 7.033936527436318e-05, | |
| "loss": 1.2432, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.16001568781253064, | |
| "grad_norm": 6.86775541305542, | |
| "learning_rate": 7.011737161553081e-05, | |
| "loss": 1.7583, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.16021178546916365, | |
| "grad_norm": 5.526086330413818, | |
| "learning_rate": 6.989553954415524e-05, | |
| "loss": 1.5891, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.16040788312579665, | |
| "grad_norm": 26.293426513671875, | |
| "learning_rate": 6.967387025977215e-05, | |
| "loss": 1.7835, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.16060398078242966, | |
| "grad_norm": 5.713183879852295, | |
| "learning_rate": 6.945236496103707e-05, | |
| "loss": 1.7831, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.16080007843906266, | |
| "grad_norm": 7.669525146484375, | |
| "learning_rate": 6.92310248457187e-05, | |
| "loss": 1.8961, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.16099617609569566, | |
| "grad_norm": 7.351653099060059, | |
| "learning_rate": 6.900985111069256e-05, | |
| "loss": 1.8533, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.16119227375232867, | |
| "grad_norm": 6.009294509887695, | |
| "learning_rate": 6.878884495193448e-05, | |
| "loss": 1.7683, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.16138837140896167, | |
| "grad_norm": 5.90309476852417, | |
| "learning_rate": 6.856800756451414e-05, | |
| "loss": 1.7174, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.16158446906559468, | |
| "grad_norm": 6.903116703033447, | |
| "learning_rate": 6.83473401425886e-05, | |
| "loss": 1.6063, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.16178056672222768, | |
| "grad_norm": 5.747986793518066, | |
| "learning_rate": 6.812684387939585e-05, | |
| "loss": 1.7352, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.16197666437886069, | |
| "grad_norm": 4.83687686920166, | |
| "learning_rate": 6.79065199672484e-05, | |
| "loss": 1.0662, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.1621727620354937, | |
| "grad_norm": 4.761837482452393, | |
| "learning_rate": 6.768636959752669e-05, | |
| "loss": 1.4324, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.16236885969212667, | |
| "grad_norm": 4.991310119628906, | |
| "learning_rate": 6.746639396067283e-05, | |
| "loss": 0.6861, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.16256495734875967, | |
| "grad_norm": 9.824933052062988, | |
| "learning_rate": 6.724659424618401e-05, | |
| "loss": 1.2527, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.16276105500539267, | |
| "grad_norm": 9.043631553649902, | |
| "learning_rate": 6.702697164260626e-05, | |
| "loss": 1.2587, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.16295715266202568, | |
| "grad_norm": 6.001374244689941, | |
| "learning_rate": 6.680752733752777e-05, | |
| "loss": 1.0727, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.16315325031865868, | |
| "grad_norm": 7.015263557434082, | |
| "learning_rate": 6.658826251757264e-05, | |
| "loss": 0.963, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.1633493479752917, | |
| "grad_norm": 16.830978393554688, | |
| "learning_rate": 6.63691783683944e-05, | |
| "loss": 1.3392, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.1635454456319247, | |
| "grad_norm": 12.247193336486816, | |
| "learning_rate": 6.61502760746697e-05, | |
| "loss": 2.2754, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.1637415432885577, | |
| "grad_norm": 3.836216449737549, | |
| "learning_rate": 6.593155682009176e-05, | |
| "loss": 2.4028, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.1639376409451907, | |
| "grad_norm": 5.886529445648193, | |
| "learning_rate": 6.571302178736404e-05, | |
| "loss": 1.0402, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.1641337386018237, | |
| "grad_norm": 5.377973556518555, | |
| "learning_rate": 6.549467215819378e-05, | |
| "loss": 1.4784, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.1643298362584567, | |
| "grad_norm": 4.867081165313721, | |
| "learning_rate": 6.527650911328585e-05, | |
| "loss": 1.9024, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.1645259339150897, | |
| "grad_norm": 8.56846809387207, | |
| "learning_rate": 6.505853383233596e-05, | |
| "loss": 1.2514, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.16472203157172272, | |
| "grad_norm": 14.424854278564453, | |
| "learning_rate": 6.484074749402467e-05, | |
| "loss": 1.2373, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.16491812922835572, | |
| "grad_norm": 5.4463067054748535, | |
| "learning_rate": 6.462315127601071e-05, | |
| "loss": 1.5188, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.16511422688498872, | |
| "grad_norm": 5.752686500549316, | |
| "learning_rate": 6.440574635492493e-05, | |
| "loss": 1.0012, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.16531032454162173, | |
| "grad_norm": 5.5835442543029785, | |
| "learning_rate": 6.418853390636364e-05, | |
| "loss": 1.8797, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.16550642219825473, | |
| "grad_norm": 8.692926406860352, | |
| "learning_rate": 6.39715151048823e-05, | |
| "loss": 1.2265, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.16570251985488774, | |
| "grad_norm": 5.524386882781982, | |
| "learning_rate": 6.375469112398946e-05, | |
| "loss": 0.7208, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.16589861751152074, | |
| "grad_norm": 5.7656145095825195, | |
| "learning_rate": 6.353806313614001e-05, | |
| "loss": 1.2716, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.16609471516815374, | |
| "grad_norm": 12.103293418884277, | |
| "learning_rate": 6.332163231272911e-05, | |
| "loss": 3.2642, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.16629081282478675, | |
| "grad_norm": 22.349411010742188, | |
| "learning_rate": 6.310539982408568e-05, | |
| "loss": 2.3537, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.16648691048141975, | |
| "grad_norm": 7.782928466796875, | |
| "learning_rate": 6.288936683946633e-05, | |
| "loss": 1.1905, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.16668300813805276, | |
| "grad_norm": 6.026412487030029, | |
| "learning_rate": 6.267353452704876e-05, | |
| "loss": 1.1583, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.16687910579468576, | |
| "grad_norm": 10.18653392791748, | |
| "learning_rate": 6.245790405392553e-05, | |
| "loss": 2.5888, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.16707520345131877, | |
| "grad_norm": 6.149682998657227, | |
| "learning_rate": 6.224247658609781e-05, | |
| "loss": 1.0514, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.16727130110795177, | |
| "grad_norm": 5.528243541717529, | |
| "learning_rate": 6.202725328846909e-05, | |
| "loss": 0.8576, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.16746739876458477, | |
| "grad_norm": 7.682480812072754, | |
| "learning_rate": 6.181223532483875e-05, | |
| "loss": 1.5959, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.16766349642121778, | |
| "grad_norm": 4.653623580932617, | |
| "learning_rate": 6.159742385789588e-05, | |
| "loss": 0.7513, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.16785959407785078, | |
| "grad_norm": 18.469745635986328, | |
| "learning_rate": 6.138282004921295e-05, | |
| "loss": 3.1673, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.1680556917344838, | |
| "grad_norm": 4.403334140777588, | |
| "learning_rate": 6.116842505923955e-05, | |
| "loss": 0.9741, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.16825178939111676, | |
| "grad_norm": 5.073941230773926, | |
| "learning_rate": 6.095424004729611e-05, | |
| "loss": 0.8638, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.16844788704774977, | |
| "grad_norm": 6.09031867980957, | |
| "learning_rate": 6.0740266171567606e-05, | |
| "loss": 0.7445, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.16864398470438277, | |
| "grad_norm": 5.235655784606934, | |
| "learning_rate": 6.0526504589097254e-05, | |
| "loss": 1.79, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.16884008236101578, | |
| "grad_norm": 5.974957466125488, | |
| "learning_rate": 6.031295645578049e-05, | |
| "loss": 2.638, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.16903618001764878, | |
| "grad_norm": 4.034956932067871, | |
| "learning_rate": 6.009962292635838e-05, | |
| "loss": 1.2214, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.16923227767428178, | |
| "grad_norm": 5.100616931915283, | |
| "learning_rate": 5.988650515441156e-05, | |
| "loss": 1.7694, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.1694283753309148, | |
| "grad_norm": 6.878417491912842, | |
| "learning_rate": 5.967360429235407e-05, | |
| "loss": 1.2678, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.1696244729875478, | |
| "grad_norm": 6.435969352722168, | |
| "learning_rate": 5.946092149142692e-05, | |
| "loss": 1.449, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.1698205706441808, | |
| "grad_norm": 5.022701740264893, | |
| "learning_rate": 5.924845790169209e-05, | |
| "loss": 0.8444, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.1700166683008138, | |
| "grad_norm": 4.173498153686523, | |
| "learning_rate": 5.903621467202608e-05, | |
| "loss": 1.089, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.1702127659574468, | |
| "grad_norm": 6.396774768829346, | |
| "learning_rate": 5.88241929501139e-05, | |
| "loss": 1.1751, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.1704088636140798, | |
| "grad_norm": 4.628440856933594, | |
| "learning_rate": 5.861239388244273e-05, | |
| "loss": 1.4271, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.1706049612707128, | |
| "grad_norm": 13.596205711364746, | |
| "learning_rate": 5.8400818614295806e-05, | |
| "loss": 0.7325, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.17080105892734582, | |
| "grad_norm": 8.975894927978516, | |
| "learning_rate": 5.818946828974607e-05, | |
| "loss": 1.8689, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.17099715658397882, | |
| "grad_norm": 5.900057315826416, | |
| "learning_rate": 5.797834405165035e-05, | |
| "loss": 0.8738, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.17119325424061183, | |
| "grad_norm": 6.28643798828125, | |
| "learning_rate": 5.776744704164272e-05, | |
| "loss": 2.4116, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.17138935189724483, | |
| "grad_norm": 6.7851033210754395, | |
| "learning_rate": 5.7556778400128583e-05, | |
| "loss": 1.5964, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.17158544955387783, | |
| "grad_norm": 12.063843727111816, | |
| "learning_rate": 5.7346339266278505e-05, | |
| "loss": 1.9227, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.17178154721051084, | |
| "grad_norm": 6.9710469245910645, | |
| "learning_rate": 5.713613077802199e-05, | |
| "loss": 1.3143, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.17197764486714384, | |
| "grad_norm": 10.355106353759766, | |
| "learning_rate": 5.6926154072041295e-05, | |
| "loss": 2.005, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.17217374252377685, | |
| "grad_norm": 8.473671913146973, | |
| "learning_rate": 5.671641028376546e-05, | |
| "loss": 1.4054, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.17236984018040985, | |
| "grad_norm": 6.026031494140625, | |
| "learning_rate": 5.650690054736396e-05, | |
| "loss": 0.9981, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.17256593783704285, | |
| "grad_norm": 7.832981109619141, | |
| "learning_rate": 5.629762599574064e-05, | |
| "loss": 1.318, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.17276203549367586, | |
| "grad_norm": 8.796941757202148, | |
| "learning_rate": 5.6088587760527656e-05, | |
| "loss": 1.7696, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.17295813315030886, | |
| "grad_norm": 5.9997124671936035, | |
| "learning_rate": 5.587978697207927e-05, | |
| "loss": 1.5515, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.17315423080694187, | |
| "grad_norm": 10.373340606689453, | |
| "learning_rate": 5.5671224759465775e-05, | |
| "loss": 1.735, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.17335032846357487, | |
| "grad_norm": 6.619256973266602, | |
| "learning_rate": 5.546290225046744e-05, | |
| "loss": 1.3402, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.17354642612020788, | |
| "grad_norm": 7.469152927398682, | |
| "learning_rate": 5.5254820571568325e-05, | |
| "loss": 1.0792, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.17374252377684088, | |
| "grad_norm": 4.927844524383545, | |
| "learning_rate": 5.504698084795014e-05, | |
| "loss": 1.4485, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.17393862143347388, | |
| "grad_norm": 4.025864601135254, | |
| "learning_rate": 5.483938420348642e-05, | |
| "loss": 0.9515, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.17413471909010686, | |
| "grad_norm": 14.059304237365723, | |
| "learning_rate": 5.4632031760736146e-05, | |
| "loss": 1.9276, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.17433081674673986, | |
| "grad_norm": 6.336188793182373, | |
| "learning_rate": 5.4424924640937824e-05, | |
| "loss": 1.7752, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.17452691440337287, | |
| "grad_norm": 7.820261478424072, | |
| "learning_rate": 5.421806396400339e-05, | |
| "loss": 1.4182, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.17472301206000587, | |
| "grad_norm": 3.8585610389709473, | |
| "learning_rate": 5.4011450848512225e-05, | |
| "loss": 1.1074, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.17491910971663888, | |
| "grad_norm": 5.7243123054504395, | |
| "learning_rate": 5.3805086411704985e-05, | |
| "loss": 2.2836, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.17511520737327188, | |
| "grad_norm": 3.6346094608306885, | |
| "learning_rate": 5.3598971769477655e-05, | |
| "loss": 1.2185, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.17531130502990488, | |
| "grad_norm": 5.283199310302734, | |
| "learning_rate": 5.339310803637538e-05, | |
| "loss": 1.0693, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.1755074026865379, | |
| "grad_norm": 5.19635009765625, | |
| "learning_rate": 5.318749632558674e-05, | |
| "loss": 1.3408, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.1757035003431709, | |
| "grad_norm": 7.379054546356201, | |
| "learning_rate": 5.298213774893735e-05, | |
| "loss": 1.6441, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.1758995979998039, | |
| "grad_norm": 10.650435447692871, | |
| "learning_rate": 5.277703341688407e-05, | |
| "loss": 1.1186, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.1760956956564369, | |
| "grad_norm": 4.219560623168945, | |
| "learning_rate": 5.257218443850895e-05, | |
| "loss": 1.1465, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.1762917933130699, | |
| "grad_norm": 9.091025352478027, | |
| "learning_rate": 5.236759192151336e-05, | |
| "loss": 2.0643, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.1764878909697029, | |
| "grad_norm": 6.650880336761475, | |
| "learning_rate": 5.2163256972211714e-05, | |
| "loss": 1.6255, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.1766839886263359, | |
| "grad_norm": 5.008049011230469, | |
| "learning_rate": 5.195918069552572e-05, | |
| "loss": 1.0065, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.17688008628296892, | |
| "grad_norm": 6.407417297363281, | |
| "learning_rate": 5.175536419497831e-05, | |
| "loss": 0.7189, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.17707618393960192, | |
| "grad_norm": 4.843297004699707, | |
| "learning_rate": 5.155180857268779e-05, | |
| "loss": 1.5639, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.17727228159623493, | |
| "grad_norm": 3.7099854946136475, | |
| "learning_rate": 5.1348514929361714e-05, | |
| "loss": 1.0337, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.17746837925286793, | |
| "grad_norm": 7.940112590789795, | |
| "learning_rate": 5.114548436429098e-05, | |
| "loss": 1.9378, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.17766447690950093, | |
| "grad_norm": 6.462250232696533, | |
| "learning_rate": 5.0942717975344035e-05, | |
| "loss": 1.2462, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.17786057456613394, | |
| "grad_norm": 3.7596821784973145, | |
| "learning_rate": 5.074021685896072e-05, | |
| "loss": 0.5025, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.17805667222276694, | |
| "grad_norm": 5.532608509063721, | |
| "learning_rate": 5.0537982110146464e-05, | |
| "loss": 1.2263, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.17825276987939995, | |
| "grad_norm": 6.286501884460449, | |
| "learning_rate": 5.033601482246628e-05, | |
| "loss": 0.9899, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.17844886753603295, | |
| "grad_norm": 8.714296340942383, | |
| "learning_rate": 5.013431608803909e-05, | |
| "loss": 1.0626, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.17864496519266596, | |
| "grad_norm": 6.827104568481445, | |
| "learning_rate": 4.993288699753146e-05, | |
| "loss": 1.1741, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.17884106284929896, | |
| "grad_norm": 7.155247688293457, | |
| "learning_rate": 4.973172864015193e-05, | |
| "loss": 1.9644, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.17903716050593196, | |
| "grad_norm": 4.7466139793396, | |
| "learning_rate": 4.953084210364508e-05, | |
| "loss": 1.2649, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.17923325816256497, | |
| "grad_norm": 6.292973041534424, | |
| "learning_rate": 4.933022847428571e-05, | |
| "loss": 2.3925, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.17942935581919797, | |
| "grad_norm": 6.9081950187683105, | |
| "learning_rate": 4.912988883687283e-05, | |
| "loss": 2.7031, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.17962545347583098, | |
| "grad_norm": 10.206450462341309, | |
| "learning_rate": 4.892982427472387e-05, | |
| "loss": 1.2479, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.17982155113246398, | |
| "grad_norm": 9.753491401672363, | |
| "learning_rate": 4.873003586966881e-05, | |
| "loss": 1.1683, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.18001764878909696, | |
| "grad_norm": 5.422173023223877, | |
| "learning_rate": 4.853052470204438e-05, | |
| "loss": 1.0117, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.18021374644572996, | |
| "grad_norm": 4.505789756774902, | |
| "learning_rate": 4.8331291850688096e-05, | |
| "loss": 0.7031, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.18040984410236297, | |
| "grad_norm": 6.756499290466309, | |
| "learning_rate": 4.813233839293265e-05, | |
| "loss": 1.5295, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.18060594175899597, | |
| "grad_norm": 7.535793781280518, | |
| "learning_rate": 4.7933665404599804e-05, | |
| "loss": 1.2856, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.18080203941562897, | |
| "grad_norm": 8.04592227935791, | |
| "learning_rate": 4.773527395999474e-05, | |
| "loss": 0.9891, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.18099813707226198, | |
| "grad_norm": 5.000694274902344, | |
| "learning_rate": 4.753716513190024e-05, | |
| "loss": 1.3128, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.18119423472889498, | |
| "grad_norm": 8.065174102783203, | |
| "learning_rate": 4.7339339991570876e-05, | |
| "loss": 1.9393, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.18139033238552799, | |
| "grad_norm": 6.1632256507873535, | |
| "learning_rate": 4.714179960872712e-05, | |
| "loss": 1.7183, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.181586430042161, | |
| "grad_norm": 6.445769786834717, | |
| "learning_rate": 4.69445450515498e-05, | |
| "loss": 2.6642, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.181782527698794, | |
| "grad_norm": 8.51395320892334, | |
| "learning_rate": 4.674757738667405e-05, | |
| "loss": 1.2191, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.181978625355427, | |
| "grad_norm": 9.159540176391602, | |
| "learning_rate": 4.6550897679183625e-05, | |
| "loss": 2.1541, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.18217472301206, | |
| "grad_norm": 7.939080715179443, | |
| "learning_rate": 4.635450699260535e-05, | |
| "loss": 1.4875, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.182370820668693, | |
| "grad_norm": 11.180910110473633, | |
| "learning_rate": 4.615840638890305e-05, | |
| "loss": 1.9817, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.182566918325326, | |
| "grad_norm": 7.014312744140625, | |
| "learning_rate": 4.596259692847198e-05, | |
| "loss": 1.2057, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.18276301598195901, | |
| "grad_norm": 3.89388108253479, | |
| "learning_rate": 4.5767079670133064e-05, | |
| "loss": 1.2204, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.18295911363859202, | |
| "grad_norm": 7.551772594451904, | |
| "learning_rate": 4.557185567112725e-05, | |
| "loss": 1.271, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.18315521129522502, | |
| "grad_norm": 5.167237758636475, | |
| "learning_rate": 4.537692598710962e-05, | |
| "loss": 2.0885, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.18335130895185803, | |
| "grad_norm": 7.082968235015869, | |
| "learning_rate": 4.5182291672143796e-05, | |
| "loss": 2.0684, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.18354740660849103, | |
| "grad_norm": 5.6844801902771, | |
| "learning_rate": 4.498795377869622e-05, | |
| "loss": 1.2413, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.18374350426512404, | |
| "grad_norm": 8.226604461669922, | |
| "learning_rate": 4.479391335763054e-05, | |
| "loss": 1.7761, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.18393960192175704, | |
| "grad_norm": 6.878251075744629, | |
| "learning_rate": 4.4600171458201735e-05, | |
| "loss": 1.4162, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.18413569957839004, | |
| "grad_norm": 6.791055202484131, | |
| "learning_rate": 4.440672912805063e-05, | |
| "loss": 1.3623, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.18433179723502305, | |
| "grad_norm": 4.67439079284668, | |
| "learning_rate": 4.4213587413198085e-05, | |
| "loss": 0.9896, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.18452789489165605, | |
| "grad_norm": 6.860576629638672, | |
| "learning_rate": 4.402074735803955e-05, | |
| "loss": 1.7911, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.18472399254828906, | |
| "grad_norm": 5.80997896194458, | |
| "learning_rate": 4.382821000533913e-05, | |
| "loss": 2.8923, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.18492009020492206, | |
| "grad_norm": 5.034128189086914, | |
| "learning_rate": 4.3635976396224186e-05, | |
| "loss": 1.7938, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.18511618786155506, | |
| "grad_norm": 4.669545650482178, | |
| "learning_rate": 4.3444047570179525e-05, | |
| "loss": 0.8484, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.18531228551818807, | |
| "grad_norm": 6.147574424743652, | |
| "learning_rate": 4.3252424565042017e-05, | |
| "loss": 1.1467, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.18550838317482107, | |
| "grad_norm": 21.214567184448242, | |
| "learning_rate": 4.306110841699471e-05, | |
| "loss": 3.0427, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.18570448083145408, | |
| "grad_norm": 5.828713893890381, | |
| "learning_rate": 4.2870100160561344e-05, | |
| "loss": 0.9022, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.18590057848808705, | |
| "grad_norm": 8.350537300109863, | |
| "learning_rate": 4.267940082860088e-05, | |
| "loss": 1.5416, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.18609667614472006, | |
| "grad_norm": 3.421091318130493, | |
| "learning_rate": 4.2489011452301705e-05, | |
| "loss": 0.7294, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.18629277380135306, | |
| "grad_norm": 5.414788246154785, | |
| "learning_rate": 4.229893306117616e-05, | |
| "loss": 0.9339, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.18648887145798607, | |
| "grad_norm": 4.0585761070251465, | |
| "learning_rate": 4.210916668305496e-05, | |
| "loss": 0.9102, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.18668496911461907, | |
| "grad_norm": 13.428049087524414, | |
| "learning_rate": 4.1919713344081704e-05, | |
| "loss": 2.287, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.18688106677125207, | |
| "grad_norm": 7.367785453796387, | |
| "learning_rate": 4.17305740687072e-05, | |
| "loss": 1.5482, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.18707716442788508, | |
| "grad_norm": 4.9446635246276855, | |
| "learning_rate": 4.154174987968398e-05, | |
| "loss": 1.5513, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.18727326208451808, | |
| "grad_norm": 6.660252094268799, | |
| "learning_rate": 4.135324179806078e-05, | |
| "loss": 1.6624, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.1874693597411511, | |
| "grad_norm": 8.531050682067871, | |
| "learning_rate": 4.11650508431771e-05, | |
| "loss": 1.58, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.1876654573977841, | |
| "grad_norm": 5.770985126495361, | |
| "learning_rate": 4.09771780326575e-05, | |
| "loss": 1.2329, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.1878615550544171, | |
| "grad_norm": 6.347046375274658, | |
| "learning_rate": 4.0789624382406264e-05, | |
| "loss": 1.3816, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.1880576527110501, | |
| "grad_norm": 10.09997272491455, | |
| "learning_rate": 4.060239090660181e-05, | |
| "loss": 1.6532, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.1882537503676831, | |
| "grad_norm": 5.894500255584717, | |
| "learning_rate": 4.041547861769126e-05, | |
| "loss": 2.0952, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.1884498480243161, | |
| "grad_norm": 3.906419038772583, | |
| "learning_rate": 4.0228888526384936e-05, | |
| "loss": 1.5555, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.1886459456809491, | |
| "grad_norm": 6.6089630126953125, | |
| "learning_rate": 4.004262164165098e-05, | |
| "loss": 2.0901, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.18884204333758212, | |
| "grad_norm": 5.344242095947266, | |
| "learning_rate": 3.985667897070974e-05, | |
| "loss": 2.2121, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.18903814099421512, | |
| "grad_norm": 3.924384117126465, | |
| "learning_rate": 3.967106151902842e-05, | |
| "loss": 2.3173, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.18923423865084812, | |
| "grad_norm": 5.392980098724365, | |
| "learning_rate": 3.948577029031566e-05, | |
| "loss": 1.3514, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.18943033630748113, | |
| "grad_norm": 5.208516597747803, | |
| "learning_rate": 3.930080628651607e-05, | |
| "loss": 1.1333, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.18962643396411413, | |
| "grad_norm": 8.4103364944458, | |
| "learning_rate": 3.9116170507804784e-05, | |
| "loss": 3.0624, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.18982253162074714, | |
| "grad_norm": 4.458950996398926, | |
| "learning_rate": 3.89318639525822e-05, | |
| "loss": 1.974, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.19001862927738014, | |
| "grad_norm": 4.55155086517334, | |
| "learning_rate": 3.874788761746836e-05, | |
| "loss": 1.2587, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.19021472693401315, | |
| "grad_norm": 3.517223834991455, | |
| "learning_rate": 3.8564242497297686e-05, | |
| "loss": 0.9581, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.19041082459064615, | |
| "grad_norm": 6.368021011352539, | |
| "learning_rate": 3.8380929585113666e-05, | |
| "loss": 1.6365, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.19060692224727915, | |
| "grad_norm": 5.153208255767822, | |
| "learning_rate": 3.819794987216333e-05, | |
| "loss": 0.9107, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.19080301990391216, | |
| "grad_norm": 9.487665176391602, | |
| "learning_rate": 3.801530434789199e-05, | |
| "loss": 2.9295, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.19099911756054516, | |
| "grad_norm": 10.716341018676758, | |
| "learning_rate": 3.783299399993781e-05, | |
| "loss": 1.9429, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.19119521521717817, | |
| "grad_norm": 4.834690093994141, | |
| "learning_rate": 3.7651019814126654e-05, | |
| "loss": 1.7383, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.19139131287381117, | |
| "grad_norm": 4.203547954559326, | |
| "learning_rate": 3.74693827744665e-05, | |
| "loss": 0.7084, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.19158741053044417, | |
| "grad_norm": 8.38505744934082, | |
| "learning_rate": 3.7288083863142274e-05, | |
| "loss": 2.7231, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.19178350818707715, | |
| "grad_norm": 6.748301029205322, | |
| "learning_rate": 3.710712406051049e-05, | |
| "loss": 1.3249, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.19197960584371015, | |
| "grad_norm": 5.358412265777588, | |
| "learning_rate": 3.692650434509404e-05, | |
| "loss": 0.7995, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.19217570350034316, | |
| "grad_norm": 3.909745693206787, | |
| "learning_rate": 3.674622569357673e-05, | |
| "loss": 2.3175, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.19237180115697616, | |
| "grad_norm": 11.555536270141602, | |
| "learning_rate": 3.6566289080798153e-05, | |
| "loss": 1.7112, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.19256789881360917, | |
| "grad_norm": 10.718825340270996, | |
| "learning_rate": 3.638669547974831e-05, | |
| "loss": 1.1124, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.19276399647024217, | |
| "grad_norm": 4.483185291290283, | |
| "learning_rate": 3.6207445861562494e-05, | |
| "loss": 1.85, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.19296009412687518, | |
| "grad_norm": 4.881377696990967, | |
| "learning_rate": 3.602854119551585e-05, | |
| "loss": 0.8783, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.19315619178350818, | |
| "grad_norm": 5.172372817993164, | |
| "learning_rate": 3.584998244901826e-05, | |
| "loss": 1.1851, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.19335228944014118, | |
| "grad_norm": 8.540918350219727, | |
| "learning_rate": 3.567177058760905e-05, | |
| "loss": 0.8138, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.1935483870967742, | |
| "grad_norm": 10.905487060546875, | |
| "learning_rate": 3.549390657495188e-05, | |
| "loss": 2.8036, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.1937444847534072, | |
| "grad_norm": 6.448824405670166, | |
| "learning_rate": 3.531639137282938e-05, | |
| "loss": 2.4083, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.1939405824100402, | |
| "grad_norm": 4.409211158752441, | |
| "learning_rate": 3.5139225941138e-05, | |
| "loss": 1.2686, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.1941366800666732, | |
| "grad_norm": 4.925105094909668, | |
| "learning_rate": 3.496241123788294e-05, | |
| "loss": 3.3955, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.1943327777233062, | |
| "grad_norm": 6.541469097137451, | |
| "learning_rate": 3.478594821917279e-05, | |
| "loss": 1.4906, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.1945288753799392, | |
| "grad_norm": 7.749048233032227, | |
| "learning_rate": 3.460983783921443e-05, | |
| "loss": 1.7907, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.1947249730365722, | |
| "grad_norm": 4.833098888397217, | |
| "learning_rate": 3.44340810503079e-05, | |
| "loss": 1.2777, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.19492107069320522, | |
| "grad_norm": 5.138740062713623, | |
| "learning_rate": 3.425867880284129e-05, | |
| "loss": 1.7267, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.19511716834983822, | |
| "grad_norm": 8.098197937011719, | |
| "learning_rate": 3.4083632045285466e-05, | |
| "loss": 1.7961, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.19531326600647123, | |
| "grad_norm": 4.534963130950928, | |
| "learning_rate": 3.390894172418903e-05, | |
| "loss": 1.773, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.19550936366310423, | |
| "grad_norm": 9.311367988586426, | |
| "learning_rate": 3.373460878417315e-05, | |
| "loss": 1.2173, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.19570546131973723, | |
| "grad_norm": 4.8790388107299805, | |
| "learning_rate": 3.356063416792662e-05, | |
| "loss": 1.3003, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.19590155897637024, | |
| "grad_norm": 4.137528419494629, | |
| "learning_rate": 3.338701881620052e-05, | |
| "loss": 1.8794, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.19609765663300324, | |
| "grad_norm": 8.082904815673828, | |
| "learning_rate": 3.321376366780326e-05, | |
| "loss": 1.3442, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.19629375428963625, | |
| "grad_norm": 7.494518280029297, | |
| "learning_rate": 3.3040869659595485e-05, | |
| "loss": 2.3671, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.19648985194626925, | |
| "grad_norm": 4.482415676116943, | |
| "learning_rate": 3.286833772648504e-05, | |
| "loss": 0.7332, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.19668594960290225, | |
| "grad_norm": 6.464245796203613, | |
| "learning_rate": 3.269616880142182e-05, | |
| "loss": 1.6453, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.19688204725953526, | |
| "grad_norm": 6.342423915863037, | |
| "learning_rate": 3.252436381539291e-05, | |
| "loss": 1.5742, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.19707814491616826, | |
| "grad_norm": 2.4557156562805176, | |
| "learning_rate": 3.235292369741729e-05, | |
| "loss": 0.3941, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.19727424257280127, | |
| "grad_norm": 6.024993419647217, | |
| "learning_rate": 3.218184937454103e-05, | |
| "loss": 1.7341, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.19747034022943427, | |
| "grad_norm": 4.793758392333984, | |
| "learning_rate": 3.2011141771832184e-05, | |
| "loss": 0.8558, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.19766643788606725, | |
| "grad_norm": 8.792879104614258, | |
| "learning_rate": 3.184080181237575e-05, | |
| "loss": 0.6646, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.19786253554270025, | |
| "grad_norm": 8.334936141967773, | |
| "learning_rate": 3.167083041726886e-05, | |
| "loss": 1.1602, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.19805863319933326, | |
| "grad_norm": 5.780764579772949, | |
| "learning_rate": 3.1501228505615534e-05, | |
| "loss": 2.342, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.19825473085596626, | |
| "grad_norm": 9.708879470825195, | |
| "learning_rate": 3.1331996994521915e-05, | |
| "loss": 0.981, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.19845082851259926, | |
| "grad_norm": 5.935221195220947, | |
| "learning_rate": 3.1163136799091186e-05, | |
| "loss": 0.8119, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.19864692616923227, | |
| "grad_norm": 9.303512573242188, | |
| "learning_rate": 3.0994648832418784e-05, | |
| "loss": 1.5852, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.19884302382586527, | |
| "grad_norm": 7.7904052734375, | |
| "learning_rate": 3.082653400558724e-05, | |
| "loss": 1.4118, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.19903912148249828, | |
| "grad_norm": 4.866517066955566, | |
| "learning_rate": 3.065879322766142e-05, | |
| "loss": 0.7066, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.19923521913913128, | |
| "grad_norm": 4.0074896812438965, | |
| "learning_rate": 3.0491427405683514e-05, | |
| "loss": 1.0785, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.19943131679576429, | |
| "grad_norm": 5.893292427062988, | |
| "learning_rate": 3.0324437444668274e-05, | |
| "loss": 2.6953, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.1996274144523973, | |
| "grad_norm": 10.129006385803223, | |
| "learning_rate": 3.0157824247597922e-05, | |
| "loss": 1.4669, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.1998235121090303, | |
| "grad_norm": 6.336552143096924, | |
| "learning_rate": 2.9991588715417383e-05, | |
| "loss": 1.9093, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.2000196097656633, | |
| "grad_norm": 7.2662482261657715, | |
| "learning_rate": 2.9825731747029394e-05, | |
| "loss": 1.3408, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.2002157074222963, | |
| "grad_norm": 4.900631427764893, | |
| "learning_rate": 2.966025423928972e-05, | |
| "loss": 1.2219, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.2004118050789293, | |
| "grad_norm": 10.661945343017578, | |
| "learning_rate": 2.9495157087002123e-05, | |
| "loss": 2.2735, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.2006079027355623, | |
| "grad_norm": 6.423792362213135, | |
| "learning_rate": 2.9330441182913694e-05, | |
| "loss": 1.362, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.2006079027355623, | |
| "eval_loss": 0.3885071277618408, | |
| "eval_runtime": 77.8128, | |
| "eval_samples_per_second": 27.605, | |
| "eval_steps_per_second": 13.802, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.20080400039219531, | |
| "grad_norm": 7.308141231536865, | |
| "learning_rate": 2.9166107417709888e-05, | |
| "loss": 1.0916, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.20100009804882832, | |
| "grad_norm": 3.7046291828155518, | |
| "learning_rate": 2.9002156680009906e-05, | |
| "loss": 1.7084, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.20119619570546132, | |
| "grad_norm": 3.8992371559143066, | |
| "learning_rate": 2.883858985636165e-05, | |
| "loss": 1.0185, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.20139229336209433, | |
| "grad_norm": 4.895712852478027, | |
| "learning_rate": 2.867540783123711e-05, | |
| "loss": 1.1252, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.20158839101872733, | |
| "grad_norm": 10.118478775024414, | |
| "learning_rate": 2.851261148702744e-05, | |
| "loss": 1.8931, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.20178448867536034, | |
| "grad_norm": 9.701197624206543, | |
| "learning_rate": 2.835020170403837e-05, | |
| "loss": 2.1721, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.20198058633199334, | |
| "grad_norm": 5.3619184494018555, | |
| "learning_rate": 2.8188179360485267e-05, | |
| "loss": 1.3209, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.20217668398862634, | |
| "grad_norm": 7.866724967956543, | |
| "learning_rate": 2.8026545332488418e-05, | |
| "loss": 1.4948, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.20237278164525935, | |
| "grad_norm": 5.446761608123779, | |
| "learning_rate": 2.7865300494068435e-05, | |
| "loss": 1.041, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.20256887930189235, | |
| "grad_norm": 8.317418098449707, | |
| "learning_rate": 2.7704445717141368e-05, | |
| "loss": 1.0779, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.20276497695852536, | |
| "grad_norm": 4.434285640716553, | |
| "learning_rate": 2.7543981871514023e-05, | |
| "loss": 1.9139, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.20296107461515836, | |
| "grad_norm": 14.01183795928955, | |
| "learning_rate": 2.7383909824879294e-05, | |
| "loss": 2.4565, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.20315717227179136, | |
| "grad_norm": 4.523947715759277, | |
| "learning_rate": 2.7224230442811526e-05, | |
| "loss": 1.7748, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.20335326992842437, | |
| "grad_norm": 10.790266990661621, | |
| "learning_rate": 2.7064944588761688e-05, | |
| "loss": 1.9152, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.20354936758505734, | |
| "grad_norm": 7.028364181518555, | |
| "learning_rate": 2.6906053124052798e-05, | |
| "loss": 1.6433, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.20374546524169035, | |
| "grad_norm": 8.6144437789917, | |
| "learning_rate": 2.674755690787526e-05, | |
| "loss": 2.3653, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.20394156289832335, | |
| "grad_norm": 4.6696343421936035, | |
| "learning_rate": 2.6589456797282253e-05, | |
| "loss": 1.0972, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.20413766055495636, | |
| "grad_norm": 7.880300998687744, | |
| "learning_rate": 2.6431753647184988e-05, | |
| "loss": 1.47, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.20433375821158936, | |
| "grad_norm": 11.923666954040527, | |
| "learning_rate": 2.6274448310348178e-05, | |
| "loss": 2.1949, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.20452985586822237, | |
| "grad_norm": 11.636192321777344, | |
| "learning_rate": 2.6117541637385402e-05, | |
| "loss": 1.3088, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.20472595352485537, | |
| "grad_norm": 4.3036885261535645, | |
| "learning_rate": 2.5961034476754487e-05, | |
| "loss": 1.5687, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.20492205118148837, | |
| "grad_norm": 4.218216419219971, | |
| "learning_rate": 2.5804927674752922e-05, | |
| "loss": 0.672, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.20511814883812138, | |
| "grad_norm": 4.2901291847229, | |
| "learning_rate": 2.564922207551337e-05, | |
| "loss": 1.048, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.20531424649475438, | |
| "grad_norm": 6.806373119354248, | |
| "learning_rate": 2.549391852099896e-05, | |
| "loss": 2.2606, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.2055103441513874, | |
| "grad_norm": 5.39298152923584, | |
| "learning_rate": 2.5339017850998802e-05, | |
| "loss": 1.5175, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.2057064418080204, | |
| "grad_norm": 7.037325859069824, | |
| "learning_rate": 2.5184520903123476e-05, | |
| "loss": 1.3757, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.2059025394646534, | |
| "grad_norm": 6.3087239265441895, | |
| "learning_rate": 2.503042851280043e-05, | |
| "loss": 1.0206, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.2060986371212864, | |
| "grad_norm": 6.11033821105957, | |
| "learning_rate": 2.4876741513269597e-05, | |
| "loss": 1.99, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.2062947347779194, | |
| "grad_norm": 5.525476932525635, | |
| "learning_rate": 2.4723460735578697e-05, | |
| "loss": 0.7599, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.2064908324345524, | |
| "grad_norm": 6.608320236206055, | |
| "learning_rate": 2.4570587008578895e-05, | |
| "loss": 1.2481, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.2066869300911854, | |
| "grad_norm": 3.596764087677002, | |
| "learning_rate": 2.4418121158920227e-05, | |
| "loss": 0.713, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.20688302774781842, | |
| "grad_norm": 7.637975692749023, | |
| "learning_rate": 2.4266064011047264e-05, | |
| "loss": 1.3772, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.20707912540445142, | |
| "grad_norm": 4.366866111755371, | |
| "learning_rate": 2.411441638719445e-05, | |
| "loss": 0.5525, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.20727522306108442, | |
| "grad_norm": 5.603202819824219, | |
| "learning_rate": 2.396317910738184e-05, | |
| "loss": 0.9035, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.20747132071771743, | |
| "grad_norm": 7.776876449584961, | |
| "learning_rate": 2.3812352989410537e-05, | |
| "loss": 1.6217, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.20766741837435043, | |
| "grad_norm": 9.07315444946289, | |
| "learning_rate": 2.366193884885841e-05, | |
| "loss": 2.2412, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.20786351603098344, | |
| "grad_norm": 7.473457336425781, | |
| "learning_rate": 2.3511937499075508e-05, | |
| "loss": 1.6256, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.20805961368761644, | |
| "grad_norm": 10.729793548583984, | |
| "learning_rate": 2.3362349751179825e-05, | |
| "loss": 1.247, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.20825571134424944, | |
| "grad_norm": 6.086622714996338, | |
| "learning_rate": 2.3213176414052738e-05, | |
| "loss": 1.613, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.20845180900088245, | |
| "grad_norm": 6.6213836669921875, | |
| "learning_rate": 2.30644182943349e-05, | |
| "loss": 0.9007, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.20864790665751545, | |
| "grad_norm": 4.5187602043151855, | |
| "learning_rate": 2.2916076196421587e-05, | |
| "loss": 1.3616, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.20884400431414846, | |
| "grad_norm": 8.176258087158203, | |
| "learning_rate": 2.276815092245851e-05, | |
| "loss": 1.1822, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.20904010197078146, | |
| "grad_norm": 6.9451494216918945, | |
| "learning_rate": 2.2620643272337426e-05, | |
| "loss": 1.2958, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.20923619962741447, | |
| "grad_norm": 4.266102313995361, | |
| "learning_rate": 2.2473554043691913e-05, | |
| "loss": 1.5274, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.20943229728404747, | |
| "grad_norm": 7.066012382507324, | |
| "learning_rate": 2.2326884031892882e-05, | |
| "loss": 1.0549, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.20962839494068045, | |
| "grad_norm": 11.074556350708008, | |
| "learning_rate": 2.21806340300444e-05, | |
| "loss": 1.3012, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.20982449259731345, | |
| "grad_norm": 5.717066287994385, | |
| "learning_rate": 2.203480482897935e-05, | |
| "loss": 1.4355, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.21002059025394645, | |
| "grad_norm": 5.230037689208984, | |
| "learning_rate": 2.188939721725526e-05, | |
| "loss": 1.7493, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.21021668791057946, | |
| "grad_norm": 4.878988265991211, | |
| "learning_rate": 2.174441198114985e-05, | |
| "loss": 0.7765, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.21041278556721246, | |
| "grad_norm": 4.619425296783447, | |
| "learning_rate": 2.159984990465691e-05, | |
| "loss": 1.8523, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.21060888322384547, | |
| "grad_norm": 4.031160354614258, | |
| "learning_rate": 2.1455711769482113e-05, | |
| "loss": 0.9848, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.21080498088047847, | |
| "grad_norm": 7.096545219421387, | |
| "learning_rate": 2.131199835503861e-05, | |
| "loss": 1.4097, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.21100107853711147, | |
| "grad_norm": 5.833085060119629, | |
| "learning_rate": 2.1168710438442953e-05, | |
| "loss": 0.9551, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.21119717619374448, | |
| "grad_norm": 5.613104343414307, | |
| "learning_rate": 2.1025848794510815e-05, | |
| "loss": 1.9112, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.21139327385037748, | |
| "grad_norm": 7.113245010375977, | |
| "learning_rate": 2.0883414195752972e-05, | |
| "loss": 1.2791, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.2115893715070105, | |
| "grad_norm": 5.34241247177124, | |
| "learning_rate": 2.0741407412370838e-05, | |
| "loss": 1.4305, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.2117854691636435, | |
| "grad_norm": 4.97748327255249, | |
| "learning_rate": 2.0599829212252553e-05, | |
| "loss": 1.0597, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.2119815668202765, | |
| "grad_norm": 6.11815881729126, | |
| "learning_rate": 2.045868036096864e-05, | |
| "loss": 1.3793, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.2121776644769095, | |
| "grad_norm": 7.822902202606201, | |
| "learning_rate": 2.031796162176811e-05, | |
| "loss": 1.7602, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.2123737621335425, | |
| "grad_norm": 5.7074875831604, | |
| "learning_rate": 2.0177673755574044e-05, | |
| "loss": 1.9033, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.2125698597901755, | |
| "grad_norm": 5.141830921173096, | |
| "learning_rate": 2.0037817520979686e-05, | |
| "loss": 0.6569, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.2127659574468085, | |
| "grad_norm": 8.139151573181152, | |
| "learning_rate": 1.9898393674244243e-05, | |
| "loss": 1.6038, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.21296205510344152, | |
| "grad_norm": 6.489431858062744, | |
| "learning_rate": 1.975940296928882e-05, | |
| "loss": 1.7581, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.21315815276007452, | |
| "grad_norm": 9.372897148132324, | |
| "learning_rate": 1.9620846157692418e-05, | |
| "loss": 2.5077, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.21335425041670752, | |
| "grad_norm": 11.757204055786133, | |
| "learning_rate": 1.9482723988687734e-05, | |
| "loss": 2.0571, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.21355034807334053, | |
| "grad_norm": 4.995403289794922, | |
| "learning_rate": 1.934503720915719e-05, | |
| "loss": 1.3159, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.21374644572997353, | |
| "grad_norm": 4.217586994171143, | |
| "learning_rate": 1.9207786563628894e-05, | |
| "loss": 0.7797, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.21394254338660654, | |
| "grad_norm": 5.487307548522949, | |
| "learning_rate": 1.907097279427258e-05, | |
| "loss": 1.2424, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.21413864104323954, | |
| "grad_norm": 5.5197014808654785, | |
| "learning_rate": 1.8934596640895607e-05, | |
| "loss": 1.6582, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.21433473869987255, | |
| "grad_norm": 9.223129272460938, | |
| "learning_rate": 1.8798658840939055e-05, | |
| "loss": 1.5407, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.21453083635650555, | |
| "grad_norm": 6.948054790496826, | |
| "learning_rate": 1.8663160129473535e-05, | |
| "loss": 0.7758, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.21472693401313855, | |
| "grad_norm": 5.582278251647949, | |
| "learning_rate": 1.8528101239195395e-05, | |
| "loss": 2.0049, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.21492303166977156, | |
| "grad_norm": 4.116588115692139, | |
| "learning_rate": 1.8393482900422644e-05, | |
| "loss": 1.8004, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.21511912932640456, | |
| "grad_norm": 6.368351936340332, | |
| "learning_rate": 1.8259305841091123e-05, | |
| "loss": 1.4293, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.21531522698303757, | |
| "grad_norm": 5.690855503082275, | |
| "learning_rate": 1.8125570786750455e-05, | |
| "loss": 1.1567, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.21551132463967054, | |
| "grad_norm": 7.080834865570068, | |
| "learning_rate": 1.7992278460560152e-05, | |
| "loss": 1.1252, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.21570742229630355, | |
| "grad_norm": 3.924699544906616, | |
| "learning_rate": 1.785942958328569e-05, | |
| "loss": 1.9084, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.21590351995293655, | |
| "grad_norm": 4.245781421661377, | |
| "learning_rate": 1.772702487329474e-05, | |
| "loss": 1.2304, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.21609961760956956, | |
| "grad_norm": 11.68691349029541, | |
| "learning_rate": 1.7595065046553085e-05, | |
| "loss": 1.0978, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.21629571526620256, | |
| "grad_norm": 5.583014011383057, | |
| "learning_rate": 1.7463550816620876e-05, | |
| "loss": 2.6622, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.21649181292283556, | |
| "grad_norm": 6.454019546508789, | |
| "learning_rate": 1.7332482894648726e-05, | |
| "loss": 1.7374, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.21668791057946857, | |
| "grad_norm": 9.824748992919922, | |
| "learning_rate": 1.7201861989373926e-05, | |
| "loss": 2.4338, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.21688400823610157, | |
| "grad_norm": 4.780369758605957, | |
| "learning_rate": 1.7071688807116525e-05, | |
| "loss": 0.9636, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.21708010589273458, | |
| "grad_norm": 6.861015796661377, | |
| "learning_rate": 1.694196405177556e-05, | |
| "loss": 1.1233, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.21727620354936758, | |
| "grad_norm": 4.883321762084961, | |
| "learning_rate": 1.681268842482523e-05, | |
| "loss": 0.8476, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.21747230120600058, | |
| "grad_norm": 4.856523513793945, | |
| "learning_rate": 1.6683862625311164e-05, | |
| "loss": 1.4524, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.2176683988626336, | |
| "grad_norm": 5.7625017166137695, | |
| "learning_rate": 1.6555487349846544e-05, | |
| "loss": 1.5877, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.2178644965192666, | |
| "grad_norm": 6.9636359214782715, | |
| "learning_rate": 1.6427563292608416e-05, | |
| "loss": 1.3804, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.2180605941758996, | |
| "grad_norm": 5.53889799118042, | |
| "learning_rate": 1.630009114533384e-05, | |
| "loss": 1.4286, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.2182566918325326, | |
| "grad_norm": 8.081591606140137, | |
| "learning_rate": 1.6173071597316335e-05, | |
| "loss": 2.5756, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.2184527894891656, | |
| "grad_norm": 9.321168899536133, | |
| "learning_rate": 1.604650533540194e-05, | |
| "loss": 2.107, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.2186488871457986, | |
| "grad_norm": 7.095883846282959, | |
| "learning_rate": 1.5920393043985593e-05, | |
| "loss": 0.9074, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.2188449848024316, | |
| "grad_norm": 10.197619438171387, | |
| "learning_rate": 1.579473540500751e-05, | |
| "loss": 1.7506, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.21904108245906462, | |
| "grad_norm": 8.685538291931152, | |
| "learning_rate": 1.5669533097949328e-05, | |
| "loss": 1.6941, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.21923718011569762, | |
| "grad_norm": 6.081611156463623, | |
| "learning_rate": 1.5544786799830568e-05, | |
| "loss": 2.3827, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.21943327777233063, | |
| "grad_norm": 4.3908162117004395, | |
| "learning_rate": 1.5420497185204873e-05, | |
| "loss": 1.1959, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.21962937542896363, | |
| "grad_norm": 13.473033905029297, | |
| "learning_rate": 1.5296664926156525e-05, | |
| "loss": 2.7042, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.21982547308559663, | |
| "grad_norm": 7.351769924163818, | |
| "learning_rate": 1.5173290692296582e-05, | |
| "loss": 2.4473, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.22002157074222964, | |
| "grad_norm": 4.283702373504639, | |
| "learning_rate": 1.5050375150759433e-05, | |
| "loss": 1.2878, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.22021766839886264, | |
| "grad_norm": 6.013007164001465, | |
| "learning_rate": 1.4927918966199094e-05, | |
| "loss": 1.7944, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.22041376605549565, | |
| "grad_norm": 6.42008638381958, | |
| "learning_rate": 1.4805922800785733e-05, | |
| "loss": 1.727, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.22060986371212865, | |
| "grad_norm": 8.836663246154785, | |
| "learning_rate": 1.4684387314201919e-05, | |
| "loss": 2.6804, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.22080596136876166, | |
| "grad_norm": 6.903241157531738, | |
| "learning_rate": 1.4563313163639192e-05, | |
| "loss": 1.1425, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.22100205902539466, | |
| "grad_norm": 25.018720626831055, | |
| "learning_rate": 1.4442701003794434e-05, | |
| "loss": 2.5687, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.22119815668202766, | |
| "grad_norm": 4.393866062164307, | |
| "learning_rate": 1.4322551486866364e-05, | |
| "loss": 1.2885, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.22139425433866064, | |
| "grad_norm": 10.157917976379395, | |
| "learning_rate": 1.4202865262552045e-05, | |
| "loss": 2.6514, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.22159035199529364, | |
| "grad_norm": 8.52134895324707, | |
| "learning_rate": 1.4083642978043277e-05, | |
| "loss": 1.5537, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.22178644965192665, | |
| "grad_norm": 7.985105991363525, | |
| "learning_rate": 1.3964885278023176e-05, | |
| "loss": 1.631, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.22198254730855965, | |
| "grad_norm": 6.039412021636963, | |
| "learning_rate": 1.3846592804662638e-05, | |
| "loss": 0.9408, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.22217864496519266, | |
| "grad_norm": 6.0496697425842285, | |
| "learning_rate": 1.3728766197616905e-05, | |
| "loss": 2.2771, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.22237474262182566, | |
| "grad_norm": 6.791964530944824, | |
| "learning_rate": 1.3611406094022095e-05, | |
| "loss": 1.5164, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.22257084027845866, | |
| "grad_norm": 5.74752950668335, | |
| "learning_rate": 1.349451312849177e-05, | |
| "loss": 1.7676, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.22276693793509167, | |
| "grad_norm": 5.087756633758545, | |
| "learning_rate": 1.337808793311346e-05, | |
| "loss": 0.8442, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.22296303559172467, | |
| "grad_norm": 6.2235426902771, | |
| "learning_rate": 1.3262131137445266e-05, | |
| "loss": 1.7659, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.22315913324835768, | |
| "grad_norm": 5.494975566864014, | |
| "learning_rate": 1.314664336851248e-05, | |
| "loss": 2.0008, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.22335523090499068, | |
| "grad_norm": 7.123453617095947, | |
| "learning_rate": 1.3031625250804191e-05, | |
| "loss": 1.3864, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.22355132856162369, | |
| "grad_norm": 4.093453407287598, | |
| "learning_rate": 1.2917077406269872e-05, | |
| "loss": 0.8456, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.2237474262182567, | |
| "grad_norm": 9.605305671691895, | |
| "learning_rate": 1.280300045431605e-05, | |
| "loss": 2.1936, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.2239435238748897, | |
| "grad_norm": 5.773606777191162, | |
| "learning_rate": 1.2689395011802896e-05, | |
| "loss": 1.3258, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.2241396215315227, | |
| "grad_norm": 5.294035911560059, | |
| "learning_rate": 1.2576261693041036e-05, | |
| "loss": 1.6611, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.2243357191881557, | |
| "grad_norm": 8.730361938476562, | |
| "learning_rate": 1.2463601109788058e-05, | |
| "loss": 2.3499, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.2245318168447887, | |
| "grad_norm": 7.3704304695129395, | |
| "learning_rate": 1.235141387124531e-05, | |
| "loss": 1.6554, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.2247279145014217, | |
| "grad_norm": 6.475876331329346, | |
| "learning_rate": 1.2239700584054514e-05, | |
| "loss": 1.5567, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.22492401215805471, | |
| "grad_norm": 5.318692207336426, | |
| "learning_rate": 1.2128461852294647e-05, | |
| "loss": 1.6012, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.22512010981468772, | |
| "grad_norm": 8.062721252441406, | |
| "learning_rate": 1.2017698277478506e-05, | |
| "loss": 1.0389, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.22531620747132072, | |
| "grad_norm": 8.7105131149292, | |
| "learning_rate": 1.1907410458549517e-05, | |
| "loss": 3.1607, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.22551230512795373, | |
| "grad_norm": 8.629892349243164, | |
| "learning_rate": 1.1797598991878512e-05, | |
| "loss": 1.8994, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.22570840278458673, | |
| "grad_norm": 8.824653625488281, | |
| "learning_rate": 1.1688264471260546e-05, | |
| "loss": 0.8091, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.22590450044121974, | |
| "grad_norm": 4.759090900421143, | |
| "learning_rate": 1.1579407487911564e-05, | |
| "loss": 1.3273, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.22610059809785274, | |
| "grad_norm": 4.78541374206543, | |
| "learning_rate": 1.1471028630465285e-05, | |
| "loss": 0.8232, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.22629669575448574, | |
| "grad_norm": 6.233590602874756, | |
| "learning_rate": 1.136312848497002e-05, | |
| "loss": 2.525, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.22649279341111875, | |
| "grad_norm": 7.085466384887695, | |
| "learning_rate": 1.1255707634885526e-05, | |
| "loss": 1.6632, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.22668889106775175, | |
| "grad_norm": 4.938376426696777, | |
| "learning_rate": 1.1148766661079768e-05, | |
| "loss": 0.6869, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.22688498872438476, | |
| "grad_norm": 9.032883644104004, | |
| "learning_rate": 1.1042306141825808e-05, | |
| "loss": 1.7692, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.22708108638101776, | |
| "grad_norm": 7.324281215667725, | |
| "learning_rate": 1.0936326652798779e-05, | |
| "loss": 1.5287, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.22727718403765074, | |
| "grad_norm": 9.270583152770996, | |
| "learning_rate": 1.083082876707262e-05, | |
| "loss": 2.5909, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.22747328169428374, | |
| "grad_norm": 8.368681907653809, | |
| "learning_rate": 1.0725813055117051e-05, | |
| "loss": 1.3548, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.22766937935091675, | |
| "grad_norm": 5.176713943481445, | |
| "learning_rate": 1.0621280084794483e-05, | |
| "loss": 0.6788, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.22786547700754975, | |
| "grad_norm": 16.383853912353516, | |
| "learning_rate": 1.0517230421357016e-05, | |
| "loss": 1.4931, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.22806157466418275, | |
| "grad_norm": 6.631777286529541, | |
| "learning_rate": 1.041366462744322e-05, | |
| "loss": 1.8152, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.22825767232081576, | |
| "grad_norm": 6.580327033996582, | |
| "learning_rate": 1.0310583263075236e-05, | |
| "loss": 1.2307, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.22845376997744876, | |
| "grad_norm": 3.8760287761688232, | |
| "learning_rate": 1.0207986885655662e-05, | |
| "loss": 1.398, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.22864986763408177, | |
| "grad_norm": 4.494516849517822, | |
| "learning_rate": 1.0105876049964658e-05, | |
| "loss": 1.9966, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.22884596529071477, | |
| "grad_norm": 4.6880879402160645, | |
| "learning_rate": 1.0004251308156776e-05, | |
| "loss": 1.7868, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.22904206294734777, | |
| "grad_norm": 4.97517204284668, | |
| "learning_rate": 9.903113209758096e-06, | |
| "loss": 0.9129, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.22923816060398078, | |
| "grad_norm": 6.1455183029174805, | |
| "learning_rate": 9.802462301663218e-06, | |
| "loss": 1.0401, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.22943425826061378, | |
| "grad_norm": 5.264155387878418, | |
| "learning_rate": 9.702299128132286e-06, | |
| "loss": 0.6803, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.2296303559172468, | |
| "grad_norm": 7.0842108726501465, | |
| "learning_rate": 9.602624230788127e-06, | |
| "loss": 1.2855, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.2298264535738798, | |
| "grad_norm": 23.354158401489258, | |
| "learning_rate": 9.503438148613208e-06, | |
| "loss": 2.12, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.2300225512305128, | |
| "grad_norm": 4.803404331207275, | |
| "learning_rate": 9.404741417946783e-06, | |
| "loss": 1.6464, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.2302186488871458, | |
| "grad_norm": 5.195870876312256, | |
| "learning_rate": 9.306534572481996e-06, | |
| "loss": 2.3011, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.2304147465437788, | |
| "grad_norm": 4.52567720413208, | |
| "learning_rate": 9.208818143262965e-06, | |
| "loss": 0.6482, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.2306108442004118, | |
| "grad_norm": 6.062211990356445, | |
| "learning_rate": 9.111592658681933e-06, | |
| "loss": 1.5717, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.2308069418570448, | |
| "grad_norm": 6.672834396362305, | |
| "learning_rate": 9.014858644476444e-06, | |
| "loss": 1.468, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.23100303951367782, | |
| "grad_norm": 8.963277816772461, | |
| "learning_rate": 8.918616623726428e-06, | |
| "loss": 1.5148, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.23119913717031082, | |
| "grad_norm": 4.698469638824463, | |
| "learning_rate": 8.822867116851396e-06, | |
| "loss": 2.3368, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.23139523482694382, | |
| "grad_norm": 7.142313003540039, | |
| "learning_rate": 8.727610641607642e-06, | |
| "loss": 2.7635, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.23159133248357683, | |
| "grad_norm": 5.125174045562744, | |
| "learning_rate": 8.632847713085502e-06, | |
| "loss": 1.3292, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.23178743014020983, | |
| "grad_norm": 5.394103527069092, | |
| "learning_rate": 8.538578843706423e-06, | |
| "loss": 1.7812, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.23198352779684284, | |
| "grad_norm": 6.396340370178223, | |
| "learning_rate": 8.444804543220308e-06, | |
| "loss": 1.4261, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.23217962545347584, | |
| "grad_norm": 10.947870254516602, | |
| "learning_rate": 8.351525318702702e-06, | |
| "loss": 0.9685, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.23237572311010884, | |
| "grad_norm": 8.984673500061035, | |
| "learning_rate": 8.25874167455214e-06, | |
| "loss": 1.0452, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.23257182076674185, | |
| "grad_norm": 9.12095832824707, | |
| "learning_rate": 8.166454112487288e-06, | |
| "loss": 1.5966, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.23276791842337485, | |
| "grad_norm": 17.694997787475586, | |
| "learning_rate": 8.074663131544313e-06, | |
| "loss": 2.1847, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.23296401608000786, | |
| "grad_norm": 7.42760705947876, | |
| "learning_rate": 7.983369228074155e-06, | |
| "loss": 2.1763, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.23316011373664083, | |
| "grad_norm": 6.491481304168701, | |
| "learning_rate": 7.892572895739913e-06, | |
| "loss": 1.6988, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.23335621139327384, | |
| "grad_norm": 8.320481300354004, | |
| "learning_rate": 7.80227462551405e-06, | |
| "loss": 1.4067, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.23355230904990684, | |
| "grad_norm": 6.234914302825928, | |
| "learning_rate": 7.712474905675837e-06, | |
| "loss": 1.675, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.23374840670653985, | |
| "grad_norm": 6.310087203979492, | |
| "learning_rate": 7.62317422180866e-06, | |
| "loss": 2.4978, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.23394450436317285, | |
| "grad_norm": 7.758471488952637, | |
| "learning_rate": 7.53437305679745e-06, | |
| "loss": 1.8757, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.23414060201980585, | |
| "grad_norm": 4.8770318031311035, | |
| "learning_rate": 7.446071890826023e-06, | |
| "loss": 1.5988, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.23433669967643886, | |
| "grad_norm": 3.4752848148345947, | |
| "learning_rate": 7.358271201374478e-06, | |
| "loss": 1.526, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.23453279733307186, | |
| "grad_norm": 4.924532890319824, | |
| "learning_rate": 7.270971463216658e-06, | |
| "loss": 0.7961, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.23472889498970487, | |
| "grad_norm": 5.1697306632995605, | |
| "learning_rate": 7.184173148417561e-06, | |
| "loss": 1.114, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.23492499264633787, | |
| "grad_norm": 6.321658134460449, | |
| "learning_rate": 7.0978767263307764e-06, | |
| "loss": 1.0523, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.23512109030297088, | |
| "grad_norm": 4.862210273742676, | |
| "learning_rate": 7.012082663595931e-06, | |
| "loss": 2.3474, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.23531718795960388, | |
| "grad_norm": 6.735289096832275, | |
| "learning_rate": 6.9267914241362585e-06, | |
| "loss": 2.2363, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.23551328561623688, | |
| "grad_norm": 6.494600772857666, | |
| "learning_rate": 6.842003469155955e-06, | |
| "loss": 1.4925, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.2357093832728699, | |
| "grad_norm": 6.873384952545166, | |
| "learning_rate": 6.757719257137807e-06, | |
| "loss": 1.7298, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.2359054809295029, | |
| "grad_norm": 11.259360313415527, | |
| "learning_rate": 6.673939243840599e-06, | |
| "loss": 1.0081, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.2361015785861359, | |
| "grad_norm": 4.887358665466309, | |
| "learning_rate": 6.590663882296788e-06, | |
| "loss": 2.1434, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.2362976762427689, | |
| "grad_norm": 4.775937557220459, | |
| "learning_rate": 6.507893622809913e-06, | |
| "loss": 0.6404, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.2364937738994019, | |
| "grad_norm": 4.322073459625244, | |
| "learning_rate": 6.425628912952242e-06, | |
| "loss": 0.9223, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.2366898715560349, | |
| "grad_norm": 5.388957500457764, | |
| "learning_rate": 6.343870197562307e-06, | |
| "loss": 1.7039, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.2368859692126679, | |
| "grad_norm": 7.30275297164917, | |
| "learning_rate": 6.262617918742586e-06, | |
| "loss": 1.7324, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.23708206686930092, | |
| "grad_norm": 6.703151702880859, | |
| "learning_rate": 6.181872515857001e-06, | |
| "loss": 2.2659, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.23727816452593392, | |
| "grad_norm": 7.984116554260254, | |
| "learning_rate": 6.10163442552858e-06, | |
| "loss": 1.5091, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.23747426218256693, | |
| "grad_norm": 6.996639728546143, | |
| "learning_rate": 6.021904081637142e-06, | |
| "loss": 1.9216, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.23767035983919993, | |
| "grad_norm": 9.186728477478027, | |
| "learning_rate": 5.942681915316894e-06, | |
| "loss": 2.6348, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.23786645749583293, | |
| "grad_norm": 12.796127319335938, | |
| "learning_rate": 5.863968354954141e-06, | |
| "loss": 1.6889, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.23806255515246594, | |
| "grad_norm": 10.490285873413086, | |
| "learning_rate": 5.7857638261849314e-06, | |
| "loss": 2.4966, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.23825865280909894, | |
| "grad_norm": 4.777184009552002, | |
| "learning_rate": 5.70806875189277e-06, | |
| "loss": 1.7654, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.23845475046573195, | |
| "grad_norm": 5.094268321990967, | |
| "learning_rate": 5.630883552206367e-06, | |
| "loss": 0.7742, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.23865084812236495, | |
| "grad_norm": 3.6413767337799072, | |
| "learning_rate": 5.554208644497283e-06, | |
| "loss": 0.7798, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.23884694577899795, | |
| "grad_norm": 11.46200942993164, | |
| "learning_rate": 5.478044443377761e-06, | |
| "loss": 1.899, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.23904304343563093, | |
| "grad_norm": 6.592775821685791, | |
| "learning_rate": 5.402391360698456e-06, | |
| "loss": 0.7878, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.23923914109226393, | |
| "grad_norm": 4.292765140533447, | |
| "learning_rate": 5.327249805546175e-06, | |
| "loss": 1.1989, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.23943523874889694, | |
| "grad_norm": 5.002085208892822, | |
| "learning_rate": 5.2526201842416965e-06, | |
| "loss": 1.0088, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.23963133640552994, | |
| "grad_norm": 8.515301704406738, | |
| "learning_rate": 5.178502900337534e-06, | |
| "loss": 2.1646, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.23982743406216295, | |
| "grad_norm": 5.089056491851807, | |
| "learning_rate": 5.104898354615861e-06, | |
| "loss": 1.0671, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.24002353171879595, | |
| "grad_norm": 4.508873462677002, | |
| "learning_rate": 5.031806945086204e-06, | |
| "loss": 2.017, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.24021962937542896, | |
| "grad_norm": 5.015408992767334, | |
| "learning_rate": 4.959229066983373e-06, | |
| "loss": 2.8555, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.24041572703206196, | |
| "grad_norm": 5.895279407501221, | |
| "learning_rate": 4.887165112765291e-06, | |
| "loss": 0.8382, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.24061182468869496, | |
| "grad_norm": 4.138760566711426, | |
| "learning_rate": 4.8156154721109345e-06, | |
| "loss": 0.9274, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 0.24080792234532797, | |
| "grad_norm": 6.234400749206543, | |
| "learning_rate": 4.74458053191813e-06, | |
| "loss": 1.9066, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.24100402000196097, | |
| "grad_norm": 6.2309441566467285, | |
| "learning_rate": 4.674060676301528e-06, | |
| "loss": 1.2137, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 0.24120011765859398, | |
| "grad_norm": 6.614995956420898, | |
| "learning_rate": 4.604056286590497e-06, | |
| "loss": 2.1118, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.24139621531522698, | |
| "grad_norm": 7.630300045013428, | |
| "learning_rate": 4.534567741327112e-06, | |
| "loss": 1.028, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 0.24159231297185998, | |
| "grad_norm": 19.701765060424805, | |
| "learning_rate": 4.465595416264012e-06, | |
| "loss": 2.3872, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.241788410628493, | |
| "grad_norm": 9.740580558776855, | |
| "learning_rate": 4.397139684362462e-06, | |
| "loss": 1.8809, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 0.241984508285126, | |
| "grad_norm": 5.803571701049805, | |
| "learning_rate": 4.329200915790288e-06, | |
| "loss": 2.1341, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.242180605941759, | |
| "grad_norm": 5.827723503112793, | |
| "learning_rate": 4.261779477919892e-06, | |
| "loss": 0.8315, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.242376703598392, | |
| "grad_norm": 11.323637008666992, | |
| "learning_rate": 4.194875735326253e-06, | |
| "loss": 3.126, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.242572801255025, | |
| "grad_norm": 6.157869815826416, | |
| "learning_rate": 4.12849004978495e-06, | |
| "loss": 1.5295, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 0.242768898911658, | |
| "grad_norm": 3.730557680130005, | |
| "learning_rate": 4.062622780270253e-06, | |
| "loss": 0.8116, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.24296499656829101, | |
| "grad_norm": 7.499046325683594, | |
| "learning_rate": 3.9972742829531225e-06, | |
| "loss": 1.5176, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 0.24316109422492402, | |
| "grad_norm": 8.390069007873535, | |
| "learning_rate": 3.932444911199318e-06, | |
| "loss": 0.7114, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.24335719188155702, | |
| "grad_norm": 3.6422150135040283, | |
| "learning_rate": 3.8681350155674315e-06, | |
| "loss": 0.8022, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 0.24355328953819003, | |
| "grad_norm": 11.20263957977295, | |
| "learning_rate": 3.8043449438071256e-06, | |
| "loss": 1.6755, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.24374938719482303, | |
| "grad_norm": 8.747391700744629, | |
| "learning_rate": 3.7410750408571005e-06, | |
| "loss": 2.4434, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 0.24394548485145603, | |
| "grad_norm": 4.528903961181641, | |
| "learning_rate": 3.678325648843306e-06, | |
| "loss": 1.1245, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.24414158250808904, | |
| "grad_norm": 3.3688809871673584, | |
| "learning_rate": 3.6160971070770654e-06, | |
| "loss": 1.2526, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.24433768016472204, | |
| "grad_norm": 7.626687049865723, | |
| "learning_rate": 3.5543897520533e-06, | |
| "loss": 2.2074, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.24453377782135505, | |
| "grad_norm": 6.176375389099121, | |
| "learning_rate": 3.4932039174486174e-06, | |
| "loss": 1.9458, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 0.24472987547798805, | |
| "grad_norm": 3.6054394245147705, | |
| "learning_rate": 3.432539934119572e-06, | |
| "loss": 1.6104, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.24492597313462103, | |
| "grad_norm": 4.255305767059326, | |
| "learning_rate": 3.372398130100851e-06, | |
| "loss": 1.4093, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 0.24512207079125403, | |
| "grad_norm": 7.934302806854248, | |
| "learning_rate": 3.3127788306035134e-06, | |
| "loss": 1.5826, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.24531816844788704, | |
| "grad_norm": 3.8543338775634766, | |
| "learning_rate": 3.253682358013244e-06, | |
| "loss": 0.5287, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 0.24551426610452004, | |
| "grad_norm": 5.766990661621094, | |
| "learning_rate": 3.1951090318885434e-06, | |
| "loss": 2.0189, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.24571036376115304, | |
| "grad_norm": 5.633528232574463, | |
| "learning_rate": 3.1370591689590777e-06, | |
| "loss": 1.1827, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 0.24590646141778605, | |
| "grad_norm": 4.339439868927002, | |
| "learning_rate": 3.0795330831239313e-06, | |
| "loss": 1.5764, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.24610255907441905, | |
| "grad_norm": 7.025827407836914, | |
| "learning_rate": 3.022531085449931e-06, | |
| "loss": 1.94, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.24629865673105206, | |
| "grad_norm": 7.727778434753418, | |
| "learning_rate": 2.966053484169917e-06, | |
| "loss": 1.6985, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.24649475438768506, | |
| "grad_norm": 4.980061054229736, | |
| "learning_rate": 2.9101005846811304e-06, | |
| "loss": 0.6095, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 0.24669085204431807, | |
| "grad_norm": 9.262031555175781, | |
| "learning_rate": 2.854672689543514e-06, | |
| "loss": 1.9517, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.24688694970095107, | |
| "grad_norm": 10.269953727722168, | |
| "learning_rate": 2.7997700984781272e-06, | |
| "loss": 1.8526, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 0.24708304735758407, | |
| "grad_norm": 5.118125915527344, | |
| "learning_rate": 2.745393108365457e-06, | |
| "loss": 1.2967, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.24727914501421708, | |
| "grad_norm": 7.1361403465271, | |
| "learning_rate": 2.6915420132439085e-06, | |
| "loss": 1.9573, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 0.24747524267085008, | |
| "grad_norm": 5.416757583618164, | |
| "learning_rate": 2.638217104308127e-06, | |
| "loss": 1.2684, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.24767134032748309, | |
| "grad_norm": 5.381096363067627, | |
| "learning_rate": 2.585418669907458e-06, | |
| "loss": 0.922, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 0.2478674379841161, | |
| "grad_norm": 4.663347244262695, | |
| "learning_rate": 2.5331469955443778e-06, | |
| "loss": 0.8759, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.2480635356407491, | |
| "grad_norm": 12.182421684265137, | |
| "learning_rate": 2.4814023638729757e-06, | |
| "loss": 0.8895, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.2482596332973821, | |
| "grad_norm": 6.225351333618164, | |
| "learning_rate": 2.430185054697409e-06, | |
| "loss": 1.8041, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.2484557309540151, | |
| "grad_norm": 9.036142349243164, | |
| "learning_rate": 2.3794953449703837e-06, | |
| "loss": 2.0539, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 0.2486518286106481, | |
| "grad_norm": 6.467651844024658, | |
| "learning_rate": 2.3293335087916314e-06, | |
| "loss": 1.5472, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.2488479262672811, | |
| "grad_norm": 7.765258312225342, | |
| "learning_rate": 2.279699817406533e-06, | |
| "loss": 1.8689, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 0.24904402392391412, | |
| "grad_norm": 9.24152946472168, | |
| "learning_rate": 2.230594539204489e-06, | |
| "loss": 1.9637, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.24924012158054712, | |
| "grad_norm": 4.302461624145508, | |
| "learning_rate": 2.1820179397176287e-06, | |
| "loss": 0.8332, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 0.24943621923718012, | |
| "grad_norm": 5.98826789855957, | |
| "learning_rate": 2.133970281619246e-06, | |
| "loss": 3.038, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.24963231689381313, | |
| "grad_norm": 5.860217094421387, | |
| "learning_rate": 2.0864518247224797e-06, | |
| "loss": 1.7476, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 0.24982841455044613, | |
| "grad_norm": 7.621738910675049, | |
| "learning_rate": 2.039462825978822e-06, | |
| "loss": 1.7148, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.2500245122070791, | |
| "grad_norm": 6.9399614334106445, | |
| "learning_rate": 1.9930035394768233e-06, | |
| "loss": 1.3601, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.25022060986371214, | |
| "grad_norm": 17.67597007751465, | |
| "learning_rate": 1.947074216440592e-06, | |
| "loss": 3.0125, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.2504167075203451, | |
| "grad_norm": 4.356191158294678, | |
| "learning_rate": 1.9016751052285953e-06, | |
| "loss": 1.0786, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 0.25061280517697815, | |
| "grad_norm": 5.882324695587158, | |
| "learning_rate": 1.8568064513321715e-06, | |
| "loss": 1.2133, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.2508089028336111, | |
| "grad_norm": 4.599434852600098, | |
| "learning_rate": 1.8124684973742534e-06, | |
| "loss": 1.3988, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 0.25100500049024416, | |
| "grad_norm": 10.744729995727539, | |
| "learning_rate": 1.768661483108136e-06, | |
| "loss": 2.4072, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.25120109814687713, | |
| "grad_norm": 4.716323375701904, | |
| "learning_rate": 1.7253856454160333e-06, | |
| "loss": 2.1812, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 0.25139719580351017, | |
| "grad_norm": 3.4088003635406494, | |
| "learning_rate": 1.682641218307901e-06, | |
| "loss": 0.8226, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.25159329346014314, | |
| "grad_norm": 5.354026794433594, | |
| "learning_rate": 1.640428432920138e-06, | |
| "loss": 1.3017, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 0.2517893911167762, | |
| "grad_norm": 4.884855270385742, | |
| "learning_rate": 1.5987475175143651e-06, | |
| "loss": 0.6766, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.25198548877340915, | |
| "grad_norm": 8.052160263061523, | |
| "learning_rate": 1.557598697476148e-06, | |
| "loss": 2.2202, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.2521815864300422, | |
| "grad_norm": 5.809813976287842, | |
| "learning_rate": 1.5169821953137875e-06, | |
| "loss": 1.1507, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.25237768408667516, | |
| "grad_norm": 5.726134300231934, | |
| "learning_rate": 1.47689823065712e-06, | |
| "loss": 1.3755, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 0.2525737817433082, | |
| "grad_norm": 6.9269022941589355, | |
| "learning_rate": 1.4373470202563855e-06, | |
| "loss": 1.9541, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.25276987939994117, | |
| "grad_norm": 6.82175874710083, | |
| "learning_rate": 1.398328777980973e-06, | |
| "loss": 0.9483, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 0.2529659770565742, | |
| "grad_norm": 5.6992058753967285, | |
| "learning_rate": 1.3598437148182652e-06, | |
| "loss": 1.8013, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.2531620747132072, | |
| "grad_norm": 6.446459770202637, | |
| "learning_rate": 1.3218920388725853e-06, | |
| "loss": 1.059, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 0.2533581723698402, | |
| "grad_norm": 5.6021223068237305, | |
| "learning_rate": 1.2844739553640073e-06, | |
| "loss": 2.3983, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.2535542700264732, | |
| "grad_norm": 5.945677757263184, | |
| "learning_rate": 1.2475896666272136e-06, | |
| "loss": 1.3391, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 0.25375036768310616, | |
| "grad_norm": 8.367269515991211, | |
| "learning_rate": 1.2112393721104843e-06, | |
| "loss": 0.9355, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.2539464653397392, | |
| "grad_norm": 4.452106952667236, | |
| "learning_rate": 1.1754232683745537e-06, | |
| "loss": 1.3174, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.25414256299637217, | |
| "grad_norm": 5.392490863800049, | |
| "learning_rate": 1.1401415490916e-06, | |
| "loss": 0.7378, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.2543386606530052, | |
| "grad_norm": 7.21303129196167, | |
| "learning_rate": 1.1053944050441245e-06, | |
| "loss": 2.9174, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 0.2545347583096382, | |
| "grad_norm": 8.784597396850586, | |
| "learning_rate": 1.0711820241240067e-06, | |
| "loss": 3.5646, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.2547308559662712, | |
| "grad_norm": 6.628826141357422, | |
| "learning_rate": 1.0375045913314063e-06, | |
| "loss": 1.2581, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 0.2549269536229042, | |
| "grad_norm": 11.671515464782715, | |
| "learning_rate": 1.0043622887738413e-06, | |
| "loss": 2.1951, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.2551230512795372, | |
| "grad_norm": 4.008133888244629, | |
| "learning_rate": 9.717552956651331e-07, | |
| "loss": 2.3149, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 0.2553191489361702, | |
| "grad_norm": 8.170031547546387, | |
| "learning_rate": 9.396837883244746e-07, | |
| "loss": 1.576, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.2555152465928032, | |
| "grad_norm": 7.534716606140137, | |
| "learning_rate": 9.081479401754966e-07, | |
| "loss": 2.1834, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 0.2557113442494362, | |
| "grad_norm": 4.619508266448975, | |
| "learning_rate": 8.771479217452694e-07, | |
| "loss": 1.6227, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 0.25590744190606923, | |
| "grad_norm": 4.464870452880859, | |
| "learning_rate": 8.466839006634364e-07, | |
| "loss": 0.6817, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.2561035395627022, | |
| "grad_norm": 7.0456743240356445, | |
| "learning_rate": 8.167560416612596e-07, | |
| "loss": 2.6702, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 0.25629963721933524, | |
| "grad_norm": 3.647512674331665, | |
| "learning_rate": 7.873645065708091e-07, | |
| "loss": 2.1032, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 0.2564957348759682, | |
| "grad_norm": 10.428240776062012, | |
| "learning_rate": 7.585094543239857e-07, | |
| "loss": 1.6973, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 0.25669183253260125, | |
| "grad_norm": 4.8686299324035645, | |
| "learning_rate": 7.301910409517221e-07, | |
| "loss": 1.8198, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 0.2568879301892342, | |
| "grad_norm": 13.215803146362305, | |
| "learning_rate": 7.024094195831277e-07, | |
| "loss": 1.791, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.25708402784586726, | |
| "grad_norm": 7.418087959289551, | |
| "learning_rate": 6.751647404446781e-07, | |
| "loss": 1.0254, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 0.25728012550250023, | |
| "grad_norm": 3.0393893718719482, | |
| "learning_rate": 6.484571508593718e-07, | |
| "loss": 0.429, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.25747622315913327, | |
| "grad_norm": 8.804031372070312, | |
| "learning_rate": 6.222867952459299e-07, | |
| "loss": 1.6825, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 0.25767232081576624, | |
| "grad_norm": 13.433954238891602, | |
| "learning_rate": 5.966538151180645e-07, | |
| "loss": 1.9529, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 0.2578684184723993, | |
| "grad_norm": 5.364668369293213, | |
| "learning_rate": 5.715583490836673e-07, | |
| "loss": 1.2584, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.25806451612903225, | |
| "grad_norm": 4.066149711608887, | |
| "learning_rate": 5.470005328440664e-07, | |
| "loss": 1.6191, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 0.2582606137856653, | |
| "grad_norm": 8.135047912597656, | |
| "learning_rate": 5.22980499193304e-07, | |
| "loss": 2.0681, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 0.25845671144229826, | |
| "grad_norm": 3.4767017364501953, | |
| "learning_rate": 4.994983780174156e-07, | |
| "loss": 1.0348, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 0.2586528090989313, | |
| "grad_norm": 6.77761173248291, | |
| "learning_rate": 4.7655429629372973e-07, | |
| "loss": 0.7133, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 0.25884890675556427, | |
| "grad_norm": 4.188867092132568, | |
| "learning_rate": 4.5414837809018007e-07, | |
| "loss": 1.6745, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.2590450044121973, | |
| "grad_norm": 3.1510398387908936, | |
| "learning_rate": 4.322807445646171e-07, | |
| "loss": 0.7392, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 0.2592411020688303, | |
| "grad_norm": 14.09803581237793, | |
| "learning_rate": 4.1095151396418617e-07, | |
| "loss": 2.5941, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 0.2594371997254633, | |
| "grad_norm": 7.8199028968811035, | |
| "learning_rate": 3.9016080162466164e-07, | |
| "loss": 1.2837, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 0.2596332973820963, | |
| "grad_norm": 4.440105438232422, | |
| "learning_rate": 3.699087199698359e-07, | |
| "loss": 1.7956, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 0.25982939503872926, | |
| "grad_norm": 7.9481048583984375, | |
| "learning_rate": 3.50195378510898e-07, | |
| "loss": 0.7979, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.2600254926953623, | |
| "grad_norm": 6.204482555389404, | |
| "learning_rate": 3.310208838458562e-07, | |
| "loss": 2.1294, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 0.26022159035199527, | |
| "grad_norm": 6.299440860748291, | |
| "learning_rate": 3.1238533965897156e-07, | |
| "loss": 1.4474, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 0.2604176880086283, | |
| "grad_norm": 5.203184604644775, | |
| "learning_rate": 2.9428884672015876e-07, | |
| "loss": 1.7164, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 0.2606137856652613, | |
| "grad_norm": 5.018675804138184, | |
| "learning_rate": 2.7673150288447526e-07, | |
| "loss": 1.0184, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 0.2608098833218943, | |
| "grad_norm": 5.618327617645264, | |
| "learning_rate": 2.597134030915771e-07, | |
| "loss": 1.4616, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.2610059809785273, | |
| "grad_norm": 10.888134002685547, | |
| "learning_rate": 2.432346393652196e-07, | |
| "loss": 2.6353, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 0.2612020786351603, | |
| "grad_norm": 4.126128196716309, | |
| "learning_rate": 2.2729530081273542e-07, | |
| "loss": 1.0003, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 0.2613981762917933, | |
| "grad_norm": 7.105251789093018, | |
| "learning_rate": 2.118954736245682e-07, | |
| "loss": 1.4611, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 0.2615942739484263, | |
| "grad_norm": 4.438936233520508, | |
| "learning_rate": 1.9703524107382855e-07, | |
| "loss": 1.0069, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 0.2617903716050593, | |
| "grad_norm": 8.27482795715332, | |
| "learning_rate": 1.8271468351579446e-07, | |
| "loss": 2.8264, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.26198646926169233, | |
| "grad_norm": 7.972108840942383, | |
| "learning_rate": 1.6893387838750053e-07, | |
| "loss": 2.4485, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 0.2621825669183253, | |
| "grad_norm": 5.920525074005127, | |
| "learning_rate": 1.5569290020736039e-07, | |
| "loss": 1.4498, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 0.26237866457495834, | |
| "grad_norm": 5.699307918548584, | |
| "learning_rate": 1.429918205746672e-07, | |
| "loss": 2.0144, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 0.2625747622315913, | |
| "grad_norm": 2.4948651790618896, | |
| "learning_rate": 1.308307081693272e-07, | |
| "loss": 0.7282, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 0.26277085988822435, | |
| "grad_norm": 9.36095142364502, | |
| "learning_rate": 1.192096287513711e-07, | |
| "loss": 2.026, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.2629669575448573, | |
| "grad_norm": 11.627699851989746, | |
| "learning_rate": 1.0812864516067667e-07, | |
| "loss": 1.6524, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 0.26316305520149036, | |
| "grad_norm": 4.287498474121094, | |
| "learning_rate": 9.758781731661337e-08, | |
| "loss": 1.1361, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 0.26335915285812334, | |
| "grad_norm": 3.8631551265716553, | |
| "learning_rate": 8.758720221768713e-08, | |
| "loss": 1.0284, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 0.26355525051475637, | |
| "grad_norm": 7.8086323738098145, | |
| "learning_rate": 7.812685394127384e-08, | |
| "loss": 1.3352, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.26375134817138934, | |
| "grad_norm": 7.071966171264648, | |
| "learning_rate": 6.920682364330855e-08, | |
| "loss": 1.783, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.2639474458280224, | |
| "grad_norm": 6.961006164550781, | |
| "learning_rate": 6.082715955800789e-08, | |
| "loss": 0.666, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 0.26414354348465535, | |
| "grad_norm": 10.395101547241211, | |
| "learning_rate": 5.298790699758138e-08, | |
| "loss": 1.6436, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 0.2643396411412884, | |
| "grad_norm": 6.384310722351074, | |
| "learning_rate": 4.5689108352053867e-08, | |
| "loss": 1.5711, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 0.26453573879792136, | |
| "grad_norm": 5.545608997344971, | |
| "learning_rate": 3.893080308898789e-08, | |
| "loss": 1.2522, | |
| "step": 1349 | |
| }, | |
| { | |
| "epoch": 0.2647318364545544, | |
| "grad_norm": 17.074024200439453, | |
| "learning_rate": 3.271302775325058e-08, | |
| "loss": 3.1042, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.26492793411118737, | |
| "grad_norm": 7.239038944244385, | |
| "learning_rate": 2.7035815966891532e-08, | |
| "loss": 1.658, | |
| "step": 1351 | |
| }, | |
| { | |
| "epoch": 0.2651240317678204, | |
| "grad_norm": 6.944170951843262, | |
| "learning_rate": 2.1899198428876333e-08, | |
| "loss": 1.5184, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 0.2653201294244534, | |
| "grad_norm": 5.548795700073242, | |
| "learning_rate": 1.730320291498666e-08, | |
| "loss": 1.6825, | |
| "step": 1353 | |
| }, | |
| { | |
| "epoch": 0.26551622708108635, | |
| "grad_norm": 8.901213645935059, | |
| "learning_rate": 1.3247854277609328e-08, | |
| "loss": 1.1894, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 0.2657123247377194, | |
| "grad_norm": 9.556551933288574, | |
| "learning_rate": 9.73317444566968e-09, | |
| "loss": 2.2414, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.26590842239435236, | |
| "grad_norm": 6.794140338897705, | |
| "learning_rate": 6.759182424453947e-09, | |
| "loss": 1.3999, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 0.2661045200509854, | |
| "grad_norm": 5.626242637634277, | |
| "learning_rate": 4.325894295553745e-09, | |
| "loss": 1.5497, | |
| "step": 1357 | |
| }, | |
| { | |
| "epoch": 0.26630061770761837, | |
| "grad_norm": 6.506687164306641, | |
| "learning_rate": 2.433323216721739e-09, | |
| "loss": 2.8993, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 0.2664967153642514, | |
| "grad_norm": 5.048938274383545, | |
| "learning_rate": 1.081479421871645e-09, | |
| "loss": 1.7226, | |
| "step": 1359 | |
| }, | |
| { | |
| "epoch": 0.2666928130208844, | |
| "grad_norm": 5.443484306335449, | |
| "learning_rate": 2.7037022096720876e-10, | |
| "loss": 0.9488, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.2668889106775174, | |
| "grad_norm": 17.070817947387695, | |
| "learning_rate": 0.0, | |
| "loss": 2.8164, | |
| "step": 1361 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1361, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 341, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.914693202093015e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |