| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.996960486322188, | |
| "eval_steps": 500, | |
| "global_step": 1230, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004052684903748734, | |
| "grad_norm": 6.00320244801405, | |
| "learning_rate": 3.2520325203252037e-07, | |
| "loss": 0.7896, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008105369807497468, | |
| "grad_norm": 6.222576045032461, | |
| "learning_rate": 6.504065040650407e-07, | |
| "loss": 0.8145, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0121580547112462, | |
| "grad_norm": 6.3147466326989194, | |
| "learning_rate": 9.75609756097561e-07, | |
| "loss": 0.8239, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.016210739614994935, | |
| "grad_norm": 6.362479645547642, | |
| "learning_rate": 1.3008130081300815e-06, | |
| "loss": 0.8322, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.020263424518743668, | |
| "grad_norm": 6.006336656685738, | |
| "learning_rate": 1.6260162601626018e-06, | |
| "loss": 0.8258, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0243161094224924, | |
| "grad_norm": 5.706377700314668, | |
| "learning_rate": 1.951219512195122e-06, | |
| "loss": 0.813, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.028368794326241134, | |
| "grad_norm": 4.478033552926158, | |
| "learning_rate": 2.2764227642276426e-06, | |
| "loss": 0.7497, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03242147922998987, | |
| "grad_norm": 4.273022325963229, | |
| "learning_rate": 2.601626016260163e-06, | |
| "loss": 0.7459, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0364741641337386, | |
| "grad_norm": 2.346862817374872, | |
| "learning_rate": 2.926829268292683e-06, | |
| "loss": 0.7117, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.040526849037487336, | |
| "grad_norm": 2.04256013559385, | |
| "learning_rate": 3.2520325203252037e-06, | |
| "loss": 0.6996, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.044579533941236066, | |
| "grad_norm": 2.054871252513307, | |
| "learning_rate": 3.577235772357724e-06, | |
| "loss": 0.709, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0486322188449848, | |
| "grad_norm": 3.967566902666531, | |
| "learning_rate": 3.902439024390244e-06, | |
| "loss": 0.7214, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.05268490374873354, | |
| "grad_norm": 4.43866792477877, | |
| "learning_rate": 4.227642276422765e-06, | |
| "loss": 0.7245, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05673758865248227, | |
| "grad_norm": 4.140820102287316, | |
| "learning_rate": 4.552845528455285e-06, | |
| "loss": 0.6955, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.060790273556231005, | |
| "grad_norm": 4.219287570767295, | |
| "learning_rate": 4.8780487804878055e-06, | |
| "loss": 0.7005, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06484295845997974, | |
| "grad_norm": 3.3947757754387284, | |
| "learning_rate": 5.203252032520326e-06, | |
| "loss": 0.6672, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06889564336372847, | |
| "grad_norm": 3.1627505323677734, | |
| "learning_rate": 5.528455284552846e-06, | |
| "loss": 0.6723, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0729483282674772, | |
| "grad_norm": 2.3604944631517184, | |
| "learning_rate": 5.853658536585366e-06, | |
| "loss": 0.643, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.07700101317122594, | |
| "grad_norm": 1.577440121832204, | |
| "learning_rate": 6.178861788617887e-06, | |
| "loss": 0.6299, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08105369807497467, | |
| "grad_norm": 1.3364920747096345, | |
| "learning_rate": 6.504065040650407e-06, | |
| "loss": 0.6187, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0851063829787234, | |
| "grad_norm": 1.5891665766536718, | |
| "learning_rate": 6.829268292682928e-06, | |
| "loss": 0.6171, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08915906788247213, | |
| "grad_norm": 1.6788652551383318, | |
| "learning_rate": 7.154471544715448e-06, | |
| "loss": 0.5911, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.09321175278622088, | |
| "grad_norm": 1.357943204554921, | |
| "learning_rate": 7.4796747967479676e-06, | |
| "loss": 0.5774, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0972644376899696, | |
| "grad_norm": 0.9279852623298133, | |
| "learning_rate": 7.804878048780489e-06, | |
| "loss": 0.5843, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.10131712259371833, | |
| "grad_norm": 0.8605735421822097, | |
| "learning_rate": 8.130081300813009e-06, | |
| "loss": 0.5844, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10536980749746708, | |
| "grad_norm": 0.9179361674649504, | |
| "learning_rate": 8.45528455284553e-06, | |
| "loss": 0.5595, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.1094224924012158, | |
| "grad_norm": 0.8383820008141235, | |
| "learning_rate": 8.78048780487805e-06, | |
| "loss": 0.5713, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11347517730496454, | |
| "grad_norm": 0.6206048120081183, | |
| "learning_rate": 9.10569105691057e-06, | |
| "loss": 0.5613, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.11752786220871327, | |
| "grad_norm": 0.7728018288184204, | |
| "learning_rate": 9.43089430894309e-06, | |
| "loss": 0.5596, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.12158054711246201, | |
| "grad_norm": 0.7302788469012543, | |
| "learning_rate": 9.756097560975611e-06, | |
| "loss": 0.5517, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12563323201621074, | |
| "grad_norm": 0.5471033014715219, | |
| "learning_rate": 1.008130081300813e-05, | |
| "loss": 0.5428, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.12968591691995948, | |
| "grad_norm": 0.6021089272335061, | |
| "learning_rate": 1.0406504065040652e-05, | |
| "loss": 0.5521, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.1337386018237082, | |
| "grad_norm": 0.6344784231578812, | |
| "learning_rate": 1.0731707317073172e-05, | |
| "loss": 0.5417, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.13779128672745694, | |
| "grad_norm": 0.56723572883801, | |
| "learning_rate": 1.1056910569105692e-05, | |
| "loss": 0.5449, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.14184397163120568, | |
| "grad_norm": 0.4657734073865772, | |
| "learning_rate": 1.1382113821138213e-05, | |
| "loss": 0.5366, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1458966565349544, | |
| "grad_norm": 0.5924544278878252, | |
| "learning_rate": 1.1707317073170731e-05, | |
| "loss": 0.5175, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.14994934143870314, | |
| "grad_norm": 0.5199187612826248, | |
| "learning_rate": 1.2032520325203254e-05, | |
| "loss": 0.5258, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.1540020263424519, | |
| "grad_norm": 0.5384256176955431, | |
| "learning_rate": 1.2357723577235774e-05, | |
| "loss": 0.5428, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1580547112462006, | |
| "grad_norm": 0.4639256455088908, | |
| "learning_rate": 1.2682926829268294e-05, | |
| "loss": 0.512, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.16210739614994935, | |
| "grad_norm": 0.5170171979655123, | |
| "learning_rate": 1.3008130081300815e-05, | |
| "loss": 0.5325, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1661600810536981, | |
| "grad_norm": 0.5371223388562233, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.5296, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.1702127659574468, | |
| "grad_norm": 0.5491139329895636, | |
| "learning_rate": 1.3658536585365855e-05, | |
| "loss": 0.5137, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.17426545086119555, | |
| "grad_norm": 0.5524853187927888, | |
| "learning_rate": 1.3983739837398376e-05, | |
| "loss": 0.5077, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.17831813576494426, | |
| "grad_norm": 0.4709350119956306, | |
| "learning_rate": 1.4308943089430896e-05, | |
| "loss": 0.5223, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.182370820668693, | |
| "grad_norm": 0.5206483719669958, | |
| "learning_rate": 1.4634146341463415e-05, | |
| "loss": 0.5068, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.18642350557244175, | |
| "grad_norm": 0.48577345306844866, | |
| "learning_rate": 1.4959349593495935e-05, | |
| "loss": 0.5111, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.19047619047619047, | |
| "grad_norm": 0.4891230906864141, | |
| "learning_rate": 1.528455284552846e-05, | |
| "loss": 0.5215, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.1945288753799392, | |
| "grad_norm": 0.47829957712835924, | |
| "learning_rate": 1.5609756097560978e-05, | |
| "loss": 0.494, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.19858156028368795, | |
| "grad_norm": 0.43476521021897646, | |
| "learning_rate": 1.5934959349593496e-05, | |
| "loss": 0.495, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.20263424518743667, | |
| "grad_norm": 0.45152960534788533, | |
| "learning_rate": 1.6260162601626018e-05, | |
| "loss": 0.4899, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2066869300911854, | |
| "grad_norm": 0.41974630204356866, | |
| "learning_rate": 1.6585365853658537e-05, | |
| "loss": 0.4968, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.21073961499493415, | |
| "grad_norm": 0.4885198795156705, | |
| "learning_rate": 1.691056910569106e-05, | |
| "loss": 0.4913, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.21479229989868287, | |
| "grad_norm": 0.43215482222207086, | |
| "learning_rate": 1.7235772357723578e-05, | |
| "loss": 0.5046, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.2188449848024316, | |
| "grad_norm": 0.4195007830465775, | |
| "learning_rate": 1.75609756097561e-05, | |
| "loss": 0.4924, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.22289766970618036, | |
| "grad_norm": 0.48294202570804967, | |
| "learning_rate": 1.788617886178862e-05, | |
| "loss": 0.4744, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22695035460992907, | |
| "grad_norm": 0.45925506011499867, | |
| "learning_rate": 1.821138211382114e-05, | |
| "loss": 0.4922, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.23100303951367782, | |
| "grad_norm": 0.49597654697371446, | |
| "learning_rate": 1.8536585365853663e-05, | |
| "loss": 0.5011, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.23505572441742653, | |
| "grad_norm": 0.5518172491735049, | |
| "learning_rate": 1.886178861788618e-05, | |
| "loss": 0.4958, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.23910840932117527, | |
| "grad_norm": 0.46933087501856996, | |
| "learning_rate": 1.91869918699187e-05, | |
| "loss": 0.4822, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.24316109422492402, | |
| "grad_norm": 0.6370106280471666, | |
| "learning_rate": 1.9512195121951222e-05, | |
| "loss": 0.4858, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.24721377912867273, | |
| "grad_norm": 0.6413543139221216, | |
| "learning_rate": 1.983739837398374e-05, | |
| "loss": 0.4953, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.2512664640324215, | |
| "grad_norm": 0.49848199355152933, | |
| "learning_rate": 2.016260162601626e-05, | |
| "loss": 0.495, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.2553191489361702, | |
| "grad_norm": 0.6181581440453321, | |
| "learning_rate": 2.048780487804878e-05, | |
| "loss": 0.4911, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.25937183383991896, | |
| "grad_norm": 0.575650670169189, | |
| "learning_rate": 2.0813008130081303e-05, | |
| "loss": 0.4922, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.2634245187436677, | |
| "grad_norm": 0.48186256960236745, | |
| "learning_rate": 2.1138211382113822e-05, | |
| "loss": 0.4784, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.2674772036474164, | |
| "grad_norm": 0.5593416642374466, | |
| "learning_rate": 2.1463414634146344e-05, | |
| "loss": 0.4735, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.27152988855116517, | |
| "grad_norm": 0.469360956163329, | |
| "learning_rate": 2.1788617886178863e-05, | |
| "loss": 0.4857, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.2755825734549139, | |
| "grad_norm": 0.6564612819669285, | |
| "learning_rate": 2.2113821138211385e-05, | |
| "loss": 0.4852, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2796352583586626, | |
| "grad_norm": 0.5301294405338582, | |
| "learning_rate": 2.2439024390243907e-05, | |
| "loss": 0.466, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.28368794326241137, | |
| "grad_norm": 0.4718931669567017, | |
| "learning_rate": 2.2764227642276426e-05, | |
| "loss": 0.4747, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2877406281661601, | |
| "grad_norm": 0.5362286043763411, | |
| "learning_rate": 2.3089430894308948e-05, | |
| "loss": 0.477, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.2917933130699088, | |
| "grad_norm": 0.6442342575400131, | |
| "learning_rate": 2.3414634146341463e-05, | |
| "loss": 0.4673, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.29584599797365757, | |
| "grad_norm": 0.6434126844562428, | |
| "learning_rate": 2.3739837398373985e-05, | |
| "loss": 0.4467, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2998986828774063, | |
| "grad_norm": 0.5933172405533134, | |
| "learning_rate": 2.4065040650406507e-05, | |
| "loss": 0.4917, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.303951367781155, | |
| "grad_norm": 0.6546357515809644, | |
| "learning_rate": 2.4390243902439026e-05, | |
| "loss": 0.4699, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3080040526849038, | |
| "grad_norm": 0.6752392880665257, | |
| "learning_rate": 2.4715447154471548e-05, | |
| "loss": 0.4726, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3120567375886525, | |
| "grad_norm": 0.8266347973066085, | |
| "learning_rate": 2.5040650406504066e-05, | |
| "loss": 0.4725, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3161094224924012, | |
| "grad_norm": 0.7472338562121598, | |
| "learning_rate": 2.536585365853659e-05, | |
| "loss": 0.4749, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.32016210739615, | |
| "grad_norm": 0.6879898377422871, | |
| "learning_rate": 2.569105691056911e-05, | |
| "loss": 0.4833, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3242147922998987, | |
| "grad_norm": 0.5583778964452478, | |
| "learning_rate": 2.601626016260163e-05, | |
| "loss": 0.4685, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3282674772036474, | |
| "grad_norm": 0.655644822887041, | |
| "learning_rate": 2.634146341463415e-05, | |
| "loss": 0.4787, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.3323201621073962, | |
| "grad_norm": 0.8447391680075238, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 0.4783, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.3363728470111449, | |
| "grad_norm": 0.6610019985577404, | |
| "learning_rate": 2.699186991869919e-05, | |
| "loss": 0.4745, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.3404255319148936, | |
| "grad_norm": 0.7693647988371816, | |
| "learning_rate": 2.731707317073171e-05, | |
| "loss": 0.4738, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.3444782168186423, | |
| "grad_norm": 0.5401483226049392, | |
| "learning_rate": 2.764227642276423e-05, | |
| "loss": 0.4564, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3485309017223911, | |
| "grad_norm": 0.5571452668617228, | |
| "learning_rate": 2.796747967479675e-05, | |
| "loss": 0.4686, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3525835866261398, | |
| "grad_norm": 0.6653097392212574, | |
| "learning_rate": 2.829268292682927e-05, | |
| "loss": 0.4736, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3566362715298885, | |
| "grad_norm": 0.8081860955171563, | |
| "learning_rate": 2.8617886178861792e-05, | |
| "loss": 0.4819, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3606889564336373, | |
| "grad_norm": 1.160086329068988, | |
| "learning_rate": 2.8943089430894314e-05, | |
| "loss": 0.4735, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.364741641337386, | |
| "grad_norm": 0.7496559500504637, | |
| "learning_rate": 2.926829268292683e-05, | |
| "loss": 0.4656, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.36879432624113473, | |
| "grad_norm": 0.8462375760492464, | |
| "learning_rate": 2.959349593495935e-05, | |
| "loss": 0.4974, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.3728470111448835, | |
| "grad_norm": 0.910891540640496, | |
| "learning_rate": 2.991869918699187e-05, | |
| "loss": 0.4576, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.3768996960486322, | |
| "grad_norm": 0.8712144727172048, | |
| "learning_rate": 3.0243902439024392e-05, | |
| "loss": 0.4646, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.38095238095238093, | |
| "grad_norm": 0.881603280804917, | |
| "learning_rate": 3.056910569105692e-05, | |
| "loss": 0.4797, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.3850050658561297, | |
| "grad_norm": 1.0255702027021072, | |
| "learning_rate": 3.089430894308943e-05, | |
| "loss": 0.4836, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.3890577507598784, | |
| "grad_norm": 0.8473041167963328, | |
| "learning_rate": 3.1219512195121955e-05, | |
| "loss": 0.4611, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.39311043566362713, | |
| "grad_norm": 0.7051079923818555, | |
| "learning_rate": 3.154471544715447e-05, | |
| "loss": 0.4633, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.3971631205673759, | |
| "grad_norm": 0.7059130642892856, | |
| "learning_rate": 3.186991869918699e-05, | |
| "loss": 0.4607, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4012158054711246, | |
| "grad_norm": 0.7531896793771049, | |
| "learning_rate": 3.2195121951219514e-05, | |
| "loss": 0.4743, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.40526849037487334, | |
| "grad_norm": 0.6422938042484045, | |
| "learning_rate": 3.2520325203252037e-05, | |
| "loss": 0.4499, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4093211752786221, | |
| "grad_norm": 0.580578984504716, | |
| "learning_rate": 3.284552845528456e-05, | |
| "loss": 0.4642, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4133738601823708, | |
| "grad_norm": 0.7585976402832986, | |
| "learning_rate": 3.3170731707317074e-05, | |
| "loss": 0.4567, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.41742654508611954, | |
| "grad_norm": 0.6902428290341162, | |
| "learning_rate": 3.3495934959349596e-05, | |
| "loss": 0.4609, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.4214792299898683, | |
| "grad_norm": 0.9422494033294441, | |
| "learning_rate": 3.382113821138212e-05, | |
| "loss": 0.4635, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.425531914893617, | |
| "grad_norm": 0.9487374297602241, | |
| "learning_rate": 3.414634146341463e-05, | |
| "loss": 0.4594, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.42958459979736574, | |
| "grad_norm": 0.7203621003226016, | |
| "learning_rate": 3.4471544715447155e-05, | |
| "loss": 0.4821, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4336372847011145, | |
| "grad_norm": 0.5654528972468549, | |
| "learning_rate": 3.479674796747968e-05, | |
| "loss": 0.4789, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.4376899696048632, | |
| "grad_norm": 0.8119324408623977, | |
| "learning_rate": 3.51219512195122e-05, | |
| "loss": 0.4577, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.44174265450861194, | |
| "grad_norm": 0.9428198706242585, | |
| "learning_rate": 3.544715447154472e-05, | |
| "loss": 0.4609, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.4457953394123607, | |
| "grad_norm": 0.8710625855004989, | |
| "learning_rate": 3.577235772357724e-05, | |
| "loss": 0.4691, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.44984802431610943, | |
| "grad_norm": 0.8033960639295589, | |
| "learning_rate": 3.609756097560976e-05, | |
| "loss": 0.4512, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.45390070921985815, | |
| "grad_norm": 0.9945112083107008, | |
| "learning_rate": 3.642276422764228e-05, | |
| "loss": 0.4793, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.4579533941236069, | |
| "grad_norm": 0.7216481210249072, | |
| "learning_rate": 3.67479674796748e-05, | |
| "loss": 0.4546, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.46200607902735563, | |
| "grad_norm": 0.9714327959723378, | |
| "learning_rate": 3.7073170731707325e-05, | |
| "loss": 0.4713, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.46605876393110435, | |
| "grad_norm": 1.0660289440840254, | |
| "learning_rate": 3.739837398373984e-05, | |
| "loss": 0.4571, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.47011144883485306, | |
| "grad_norm": 0.6994984848699469, | |
| "learning_rate": 3.772357723577236e-05, | |
| "loss": 0.4617, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.47416413373860183, | |
| "grad_norm": 1.3253313131970752, | |
| "learning_rate": 3.804878048780488e-05, | |
| "loss": 0.4726, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.47821681864235055, | |
| "grad_norm": 0.7386836530128937, | |
| "learning_rate": 3.83739837398374e-05, | |
| "loss": 0.4576, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.48226950354609927, | |
| "grad_norm": 0.9939793257895867, | |
| "learning_rate": 3.869918699186992e-05, | |
| "loss": 0.4611, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.48632218844984804, | |
| "grad_norm": 0.9806219024393532, | |
| "learning_rate": 3.9024390243902444e-05, | |
| "loss": 0.4801, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.49037487335359675, | |
| "grad_norm": 0.9658328193530229, | |
| "learning_rate": 3.9349593495934966e-05, | |
| "loss": 0.4497, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.49442755825734547, | |
| "grad_norm": 1.2586735355695522, | |
| "learning_rate": 3.967479674796748e-05, | |
| "loss": 0.4691, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.49848024316109424, | |
| "grad_norm": 0.8867071239821258, | |
| "learning_rate": 4e-05, | |
| "loss": 0.4609, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.502532928064843, | |
| "grad_norm": 0.6138271894161924, | |
| "learning_rate": 3.999991946137476e-05, | |
| "loss": 0.4717, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5065856129685917, | |
| "grad_norm": 0.7351946115644556, | |
| "learning_rate": 3.999967784614766e-05, | |
| "loss": 0.4605, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5106382978723404, | |
| "grad_norm": 0.6221924332885571, | |
| "learning_rate": 3.9999275156264656e-05, | |
| "loss": 0.4499, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5146909827760892, | |
| "grad_norm": 0.587742713406424, | |
| "learning_rate": 3.999871139496895e-05, | |
| "loss": 0.4351, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5187436676798379, | |
| "grad_norm": 0.6396112304564021, | |
| "learning_rate": 3.9997986566800995e-05, | |
| "loss": 0.4654, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5227963525835866, | |
| "grad_norm": 0.6011598048018734, | |
| "learning_rate": 3.999710067759846e-05, | |
| "loss": 0.4736, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5268490374873354, | |
| "grad_norm": 0.6777202407961708, | |
| "learning_rate": 3.999605373449617e-05, | |
| "loss": 0.4653, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5309017223910841, | |
| "grad_norm": 0.5639155129382611, | |
| "learning_rate": 3.9994845745926075e-05, | |
| "loss": 0.465, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5349544072948328, | |
| "grad_norm": 0.6516840240273188, | |
| "learning_rate": 3.999347672161713e-05, | |
| "loss": 0.4617, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5390070921985816, | |
| "grad_norm": 0.6566116555266692, | |
| "learning_rate": 3.999194667259528e-05, | |
| "loss": 0.4535, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5430597771023303, | |
| "grad_norm": 0.5481081366907883, | |
| "learning_rate": 3.999025561118334e-05, | |
| "loss": 0.4567, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.547112462006079, | |
| "grad_norm": 0.8033541317296244, | |
| "learning_rate": 3.998840355100086e-05, | |
| "loss": 0.4553, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5511651469098278, | |
| "grad_norm": 0.5859366379567652, | |
| "learning_rate": 3.998639050696409e-05, | |
| "loss": 0.4502, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.5552178318135765, | |
| "grad_norm": 0.7791689056639987, | |
| "learning_rate": 3.998421649528582e-05, | |
| "loss": 0.4636, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5592705167173252, | |
| "grad_norm": 0.6460733304708896, | |
| "learning_rate": 3.9981881533475234e-05, | |
| "loss": 0.4484, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.563323201621074, | |
| "grad_norm": 0.6660274990088378, | |
| "learning_rate": 3.997938564033779e-05, | |
| "loss": 0.4514, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5673758865248227, | |
| "grad_norm": 0.7907370493144366, | |
| "learning_rate": 3.9976728835975064e-05, | |
| "loss": 0.4713, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 0.8066976394918469, | |
| "learning_rate": 3.9973911141784605e-05, | |
| "loss": 0.4598, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5754812563323202, | |
| "grad_norm": 1.050055295530242, | |
| "learning_rate": 3.997093258045973e-05, | |
| "loss": 0.476, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.5795339412360689, | |
| "grad_norm": 0.6200610643396933, | |
| "learning_rate": 3.996779317598936e-05, | |
| "loss": 0.4637, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.5835866261398176, | |
| "grad_norm": 0.7743553410356002, | |
| "learning_rate": 3.996449295365782e-05, | |
| "loss": 0.4605, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5876393110435664, | |
| "grad_norm": 0.6192635923333751, | |
| "learning_rate": 3.996103194004467e-05, | |
| "loss": 0.4476, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5916919959473151, | |
| "grad_norm": 0.6372276645436169, | |
| "learning_rate": 3.995741016302441e-05, | |
| "loss": 0.4721, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5957446808510638, | |
| "grad_norm": 0.47488295037095696, | |
| "learning_rate": 3.9953627651766364e-05, | |
| "loss": 0.4363, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.5997973657548126, | |
| "grad_norm": 0.5526106991931745, | |
| "learning_rate": 3.9949684436734325e-05, | |
| "loss": 0.4603, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6038500506585613, | |
| "grad_norm": 0.6103330649876105, | |
| "learning_rate": 3.994558054968643e-05, | |
| "loss": 0.438, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.60790273556231, | |
| "grad_norm": 0.6360809370932107, | |
| "learning_rate": 3.994131602367481e-05, | |
| "loss": 0.4676, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6119554204660588, | |
| "grad_norm": 0.7580372745147166, | |
| "learning_rate": 3.9936890893045376e-05, | |
| "loss": 0.461, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6160081053698075, | |
| "grad_norm": 0.5307352507354853, | |
| "learning_rate": 3.993230519343752e-05, | |
| "loss": 0.4616, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6200607902735562, | |
| "grad_norm": 0.6951122847454525, | |
| "learning_rate": 3.992755896178383e-05, | |
| "loss": 0.45, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.624113475177305, | |
| "grad_norm": 0.7713483823655864, | |
| "learning_rate": 3.992265223630981e-05, | |
| "loss": 0.4478, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6281661600810537, | |
| "grad_norm": 0.8056407445066684, | |
| "learning_rate": 3.991758505653355e-05, | |
| "loss": 0.4377, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6322188449848024, | |
| "grad_norm": 0.670458395877912, | |
| "learning_rate": 3.991235746326543e-05, | |
| "loss": 0.4296, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6362715298885512, | |
| "grad_norm": 0.50475319147271, | |
| "learning_rate": 3.9906969498607745e-05, | |
| "loss": 0.464, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.6403242147923, | |
| "grad_norm": 0.49040248424075855, | |
| "learning_rate": 3.990142120595444e-05, | |
| "loss": 0.4529, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6443768996960486, | |
| "grad_norm": 0.5697921850784273, | |
| "learning_rate": 3.98957126299907e-05, | |
| "loss": 0.4336, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6484295845997974, | |
| "grad_norm": 0.521708639762235, | |
| "learning_rate": 3.9889843816692596e-05, | |
| "loss": 0.4457, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6524822695035462, | |
| "grad_norm": 0.39792850953926323, | |
| "learning_rate": 3.9883814813326766e-05, | |
| "loss": 0.4418, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6565349544072948, | |
| "grad_norm": 0.4949299240248974, | |
| "learning_rate": 3.9877625668449956e-05, | |
| "loss": 0.4591, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.6605876393110436, | |
| "grad_norm": 0.604097649406502, | |
| "learning_rate": 3.98712764319087e-05, | |
| "loss": 0.4538, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.6646403242147924, | |
| "grad_norm": 0.4918604742080202, | |
| "learning_rate": 3.9864767154838864e-05, | |
| "loss": 0.4345, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.668693009118541, | |
| "grad_norm": 0.4570418745552427, | |
| "learning_rate": 3.9858097889665277e-05, | |
| "loss": 0.4516, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6727456940222898, | |
| "grad_norm": 0.6148547598738403, | |
| "learning_rate": 3.985126869010129e-05, | |
| "loss": 0.4489, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6767983789260384, | |
| "grad_norm": 0.5360329944191708, | |
| "learning_rate": 3.984427961114833e-05, | |
| "loss": 0.4545, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.6808510638297872, | |
| "grad_norm": 0.47844863049227115, | |
| "learning_rate": 3.9837130709095475e-05, | |
| "loss": 0.4307, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.684903748733536, | |
| "grad_norm": 0.5442191413314459, | |
| "learning_rate": 3.982982204151901e-05, | |
| "loss": 0.4533, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6889564336372846, | |
| "grad_norm": 0.5689267305455523, | |
| "learning_rate": 3.982235366728193e-05, | |
| "loss": 0.4468, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6930091185410334, | |
| "grad_norm": 0.536308029641486, | |
| "learning_rate": 3.9814725646533505e-05, | |
| "loss": 0.4566, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6970618034447822, | |
| "grad_norm": 0.520215037632852, | |
| "learning_rate": 3.9806938040708746e-05, | |
| "loss": 0.4489, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7011144883485309, | |
| "grad_norm": 0.48763000753278635, | |
| "learning_rate": 3.9798990912527976e-05, | |
| "loss": 0.4604, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7051671732522796, | |
| "grad_norm": 0.4806877369331677, | |
| "learning_rate": 3.979088432599627e-05, | |
| "loss": 0.4545, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7092198581560284, | |
| "grad_norm": 0.5086648309609434, | |
| "learning_rate": 3.9782618346402964e-05, | |
| "loss": 0.4429, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.713272543059777, | |
| "grad_norm": 0.4557145361712599, | |
| "learning_rate": 3.977419304032111e-05, | |
| "loss": 0.4468, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7173252279635258, | |
| "grad_norm": 0.4770606760402681, | |
| "learning_rate": 3.976560847560697e-05, | |
| "loss": 0.4488, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7213779128672746, | |
| "grad_norm": 0.45616757829814647, | |
| "learning_rate": 3.9756864721399456e-05, | |
| "loss": 0.447, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7254305977710233, | |
| "grad_norm": 0.46343336242437516, | |
| "learning_rate": 3.974796184811956e-05, | |
| "loss": 0.4423, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.729483282674772, | |
| "grad_norm": 0.48491101165923955, | |
| "learning_rate": 3.973889992746979e-05, | |
| "loss": 0.4572, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7335359675785208, | |
| "grad_norm": 0.5644298300990013, | |
| "learning_rate": 3.972967903243361e-05, | |
| "loss": 0.4436, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7375886524822695, | |
| "grad_norm": 0.40733898495588644, | |
| "learning_rate": 3.972029923727486e-05, | |
| "loss": 0.4516, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7416413373860182, | |
| "grad_norm": 0.45191897568465433, | |
| "learning_rate": 3.971076061753709e-05, | |
| "loss": 0.4551, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.745694022289767, | |
| "grad_norm": 0.4484113252969927, | |
| "learning_rate": 3.9701063250043066e-05, | |
| "loss": 0.4388, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7497467071935157, | |
| "grad_norm": 0.4429520470392494, | |
| "learning_rate": 3.969120721289402e-05, | |
| "loss": 0.4574, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7537993920972644, | |
| "grad_norm": 0.4167740678331145, | |
| "learning_rate": 3.9681192585469146e-05, | |
| "loss": 0.4306, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.7578520770010132, | |
| "grad_norm": 0.39594018532944536, | |
| "learning_rate": 3.9671019448424865e-05, | |
| "loss": 0.4377, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.7619047619047619, | |
| "grad_norm": 0.42235405329718484, | |
| "learning_rate": 3.966068788369422e-05, | |
| "loss": 0.4429, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.7659574468085106, | |
| "grad_norm": 0.43860894445430154, | |
| "learning_rate": 3.965019797448622e-05, | |
| "loss": 0.4389, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.7700101317122594, | |
| "grad_norm": 0.5100253614184337, | |
| "learning_rate": 3.963954980528515e-05, | |
| "loss": 0.4657, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.7740628166160081, | |
| "grad_norm": 0.5800829995821355, | |
| "learning_rate": 3.9628743461849905e-05, | |
| "loss": 0.4468, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.7781155015197568, | |
| "grad_norm": 0.5472809838138357, | |
| "learning_rate": 3.961777903121329e-05, | |
| "loss": 0.4584, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.7821681864235056, | |
| "grad_norm": 0.47066784587358523, | |
| "learning_rate": 3.960665660168131e-05, | |
| "loss": 0.4318, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7862208713272543, | |
| "grad_norm": 0.5866230630948102, | |
| "learning_rate": 3.9595376262832485e-05, | |
| "loss": 0.4424, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.790273556231003, | |
| "grad_norm": 0.651217473103192, | |
| "learning_rate": 3.9583938105517127e-05, | |
| "loss": 0.4608, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.7943262411347518, | |
| "grad_norm": 0.6505271921832835, | |
| "learning_rate": 3.957234222185657e-05, | |
| "loss": 0.4352, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.7983789260385005, | |
| "grad_norm": 0.44427677493615747, | |
| "learning_rate": 3.9560588705242474e-05, | |
| "loss": 0.4425, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8024316109422492, | |
| "grad_norm": 0.6477702817909823, | |
| "learning_rate": 3.954867765033605e-05, | |
| "loss": 0.4635, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.806484295845998, | |
| "grad_norm": 0.5952165902799125, | |
| "learning_rate": 3.953660915306728e-05, | |
| "loss": 0.4451, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8105369807497467, | |
| "grad_norm": 0.4651900836835382, | |
| "learning_rate": 3.952438331063419e-05, | |
| "loss": 0.4178, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8145896656534954, | |
| "grad_norm": 0.5723497989630817, | |
| "learning_rate": 3.951200022150205e-05, | |
| "loss": 0.4534, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.8186423505572442, | |
| "grad_norm": 0.7278533211932171, | |
| "learning_rate": 3.949945998540253e-05, | |
| "loss": 0.4386, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.8226950354609929, | |
| "grad_norm": 0.5049623805734051, | |
| "learning_rate": 3.9486762703332993e-05, | |
| "loss": 0.4346, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.8267477203647416, | |
| "grad_norm": 0.5755978625834166, | |
| "learning_rate": 3.947390847755559e-05, | |
| "loss": 0.4595, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.8308004052684904, | |
| "grad_norm": 0.5632901966067324, | |
| "learning_rate": 3.946089741159648e-05, | |
| "loss": 0.4444, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.8348530901722391, | |
| "grad_norm": 0.5772454428463247, | |
| "learning_rate": 3.944772961024501e-05, | |
| "loss": 0.4481, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.8389057750759878, | |
| "grad_norm": 0.6321852614754302, | |
| "learning_rate": 3.943440517955285e-05, | |
| "loss": 0.4366, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.8429584599797366, | |
| "grad_norm": 0.4630921213456246, | |
| "learning_rate": 3.9420924226833126e-05, | |
| "loss": 0.4579, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.8470111448834853, | |
| "grad_norm": 0.4435588971396196, | |
| "learning_rate": 3.9407286860659566e-05, | |
| "loss": 0.4404, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.851063829787234, | |
| "grad_norm": 0.550586168771183, | |
| "learning_rate": 3.9393493190865657e-05, | |
| "loss": 0.4628, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.8551165146909828, | |
| "grad_norm": 0.49681276675490504, | |
| "learning_rate": 3.937954332854371e-05, | |
| "loss": 0.4509, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.8591691995947315, | |
| "grad_norm": 0.4182948479525834, | |
| "learning_rate": 3.9365437386044016e-05, | |
| "loss": 0.4508, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.8632218844984803, | |
| "grad_norm": 0.45212762448571214, | |
| "learning_rate": 3.935117547697387e-05, | |
| "loss": 0.449, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.867274569402229, | |
| "grad_norm": 0.42607512114987356, | |
| "learning_rate": 3.933675771619675e-05, | |
| "loss": 0.4411, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.8713272543059777, | |
| "grad_norm": 0.3794789696396441, | |
| "learning_rate": 3.932218421983131e-05, | |
| "loss": 0.4387, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.8753799392097265, | |
| "grad_norm": 0.42905608768631476, | |
| "learning_rate": 3.9307455105250484e-05, | |
| "loss": 0.4426, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.8794326241134752, | |
| "grad_norm": 0.4039142553183988, | |
| "learning_rate": 3.929257049108054e-05, | |
| "loss": 0.438, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.8834853090172239, | |
| "grad_norm": 0.425649450863724, | |
| "learning_rate": 3.927753049720011e-05, | |
| "loss": 0.4516, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.8875379939209727, | |
| "grad_norm": 0.45528677832305864, | |
| "learning_rate": 3.9262335244739234e-05, | |
| "loss": 0.4328, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.8915906788247214, | |
| "grad_norm": 0.4286114826156143, | |
| "learning_rate": 3.92469848560784e-05, | |
| "loss": 0.4556, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.8956433637284701, | |
| "grad_norm": 0.5094241764451155, | |
| "learning_rate": 3.923147945484751e-05, | |
| "loss": 0.4371, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.8996960486322189, | |
| "grad_norm": 0.50929861386035, | |
| "learning_rate": 3.9215819165924956e-05, | |
| "loss": 0.4421, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.9037487335359676, | |
| "grad_norm": 0.5008741478998536, | |
| "learning_rate": 3.920000411543654e-05, | |
| "loss": 0.4595, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.9078014184397163, | |
| "grad_norm": 0.508314272630767, | |
| "learning_rate": 3.9184034430754495e-05, | |
| "loss": 0.437, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.9118541033434651, | |
| "grad_norm": 0.5477484359240532, | |
| "learning_rate": 3.916791024049648e-05, | |
| "loss": 0.4331, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.9159067882472138, | |
| "grad_norm": 0.4256357416859591, | |
| "learning_rate": 3.91516316745245e-05, | |
| "loss": 0.4409, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.9199594731509625, | |
| "grad_norm": 0.5189828635114462, | |
| "learning_rate": 3.913519886394389e-05, | |
| "loss": 0.4449, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.9240121580547113, | |
| "grad_norm": 0.49997896846934553, | |
| "learning_rate": 3.911861194110225e-05, | |
| "loss": 0.4383, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.92806484295846, | |
| "grad_norm": 0.5129220329357647, | |
| "learning_rate": 3.910187103958837e-05, | |
| "loss": 0.4465, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.9321175278622087, | |
| "grad_norm": 0.5822017736209879, | |
| "learning_rate": 3.908497629423117e-05, | |
| "loss": 0.4587, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.9361702127659575, | |
| "grad_norm": 0.3827715630152086, | |
| "learning_rate": 3.9067927841098614e-05, | |
| "loss": 0.4422, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.9402228976697061, | |
| "grad_norm": 0.5105251199875508, | |
| "learning_rate": 3.9050725817496594e-05, | |
| "loss": 0.4436, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.9442755825734549, | |
| "grad_norm": 0.49580699001872186, | |
| "learning_rate": 3.9033370361967844e-05, | |
| "loss": 0.4274, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.9483282674772037, | |
| "grad_norm": 0.4331567776063567, | |
| "learning_rate": 3.901586161429081e-05, | |
| "loss": 0.4357, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 0.5247819285334661, | |
| "learning_rate": 3.8998199715478545e-05, | |
| "loss": 0.445, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.9564336372847011, | |
| "grad_norm": 0.41494275749233606, | |
| "learning_rate": 3.8980384807777564e-05, | |
| "loss": 0.4512, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.9604863221884499, | |
| "grad_norm": 0.44862699306389814, | |
| "learning_rate": 3.896241703466667e-05, | |
| "loss": 0.4498, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.9645390070921985, | |
| "grad_norm": 0.44825324416458767, | |
| "learning_rate": 3.894429654085585e-05, | |
| "loss": 0.4427, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.9685916919959473, | |
| "grad_norm": 0.39941582801992576, | |
| "learning_rate": 3.892602347228505e-05, | |
| "loss": 0.4389, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.9726443768996961, | |
| "grad_norm": 0.5043657738385401, | |
| "learning_rate": 3.890759797612307e-05, | |
| "loss": 0.4123, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.9766970618034447, | |
| "grad_norm": 0.4195328837032882, | |
| "learning_rate": 3.888902020076632e-05, | |
| "loss": 0.4388, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.9807497467071935, | |
| "grad_norm": 0.45546572803631213, | |
| "learning_rate": 3.887029029583764e-05, | |
| "loss": 0.4476, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.9848024316109423, | |
| "grad_norm": 0.47656460346991847, | |
| "learning_rate": 3.8851408412185125e-05, | |
| "loss": 0.4357, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.9888551165146909, | |
| "grad_norm": 0.47802051199261913, | |
| "learning_rate": 3.8832374701880855e-05, | |
| "loss": 0.4266, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.9929078014184397, | |
| "grad_norm": 0.4689747958800069, | |
| "learning_rate": 3.881318931821972e-05, | |
| "loss": 0.4412, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.9969604863221885, | |
| "grad_norm": 0.49765756932045974, | |
| "learning_rate": 3.879385241571817e-05, | |
| "loss": 0.4386, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.0030395136778116, | |
| "grad_norm": 0.46564475330812416, | |
| "learning_rate": 3.8774364150112955e-05, | |
| "loss": 0.4136, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.0070921985815602, | |
| "grad_norm": 0.4652733117839491, | |
| "learning_rate": 3.8754724678359884e-05, | |
| "loss": 0.3895, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.011144883485309, | |
| "grad_norm": 0.4725243837515156, | |
| "learning_rate": 3.873493415863256e-05, | |
| "loss": 0.3928, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.0151975683890577, | |
| "grad_norm": 0.40681633537574413, | |
| "learning_rate": 3.871499275032111e-05, | |
| "loss": 0.3849, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.0192502532928065, | |
| "grad_norm": 0.43554767495437385, | |
| "learning_rate": 3.869490061403091e-05, | |
| "loss": 0.4006, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.0233029381965553, | |
| "grad_norm": 0.4684737848776916, | |
| "learning_rate": 3.867465791158124e-05, | |
| "loss": 0.3844, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.027355623100304, | |
| "grad_norm": 0.4271599435616579, | |
| "learning_rate": 3.865426480600407e-05, | |
| "loss": 0.4026, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.0314083080040526, | |
| "grad_norm": 0.42731644251052026, | |
| "learning_rate": 3.863372146154264e-05, | |
| "loss": 0.3895, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.0354609929078014, | |
| "grad_norm": 0.48546634483114126, | |
| "learning_rate": 3.861302804365024e-05, | |
| "loss": 0.3749, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.0395136778115501, | |
| "grad_norm": 0.4585159974079783, | |
| "learning_rate": 3.85921847189888e-05, | |
| "loss": 0.3875, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.043566362715299, | |
| "grad_norm": 0.5729812618366336, | |
| "learning_rate": 3.85711916554276e-05, | |
| "loss": 0.3979, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.0476190476190477, | |
| "grad_norm": 0.47798618608137305, | |
| "learning_rate": 3.85500490220419e-05, | |
| "loss": 0.3965, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.0516717325227964, | |
| "grad_norm": 0.6573669889689361, | |
| "learning_rate": 3.852875698911154e-05, | |
| "loss": 0.384, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.055724417426545, | |
| "grad_norm": 0.45733819922040536, | |
| "learning_rate": 3.850731572811963e-05, | |
| "loss": 0.4054, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.0597771023302938, | |
| "grad_norm": 0.4515682782605001, | |
| "learning_rate": 3.848572541175116e-05, | |
| "loss": 0.3904, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.0638297872340425, | |
| "grad_norm": 0.40206020874684595, | |
| "learning_rate": 3.846398621389154e-05, | |
| "loss": 0.3872, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.0678824721377913, | |
| "grad_norm": 0.40620032857455046, | |
| "learning_rate": 3.84420983096253e-05, | |
| "loss": 0.3942, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.07193515704154, | |
| "grad_norm": 0.41450531456591755, | |
| "learning_rate": 3.8420061875234606e-05, | |
| "loss": 0.3861, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.0759878419452888, | |
| "grad_norm": 0.4075276953014285, | |
| "learning_rate": 3.839787708819787e-05, | |
| "loss": 0.3998, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.0800405268490374, | |
| "grad_norm": 0.4331190973962223, | |
| "learning_rate": 3.8375544127188325e-05, | |
| "loss": 0.3878, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.0840932117527862, | |
| "grad_norm": 0.46282504060804075, | |
| "learning_rate": 3.8353063172072564e-05, | |
| "loss": 0.3691, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.088145896656535, | |
| "grad_norm": 0.5230390110645073, | |
| "learning_rate": 3.8330434403909105e-05, | |
| "loss": 0.3798, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.0921985815602837, | |
| "grad_norm": 0.49961966207851366, | |
| "learning_rate": 3.8307658004946934e-05, | |
| "loss": 0.3893, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.0962512664640325, | |
| "grad_norm": 0.5232700315176604, | |
| "learning_rate": 3.8284734158624046e-05, | |
| "loss": 0.3861, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.1003039513677813, | |
| "grad_norm": 0.46434641707563223, | |
| "learning_rate": 3.826166304956594e-05, | |
| "loss": 0.3953, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.1043566362715298, | |
| "grad_norm": 0.5088801400476153, | |
| "learning_rate": 3.8238444863584164e-05, | |
| "loss": 0.3797, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.1084093211752786, | |
| "grad_norm": 0.4787804663911086, | |
| "learning_rate": 3.821507978767479e-05, | |
| "loss": 0.3844, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.1124620060790273, | |
| "grad_norm": 0.6027551934834764, | |
| "learning_rate": 3.819156801001693e-05, | |
| "loss": 0.396, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.1165146909827761, | |
| "grad_norm": 0.5603169233247453, | |
| "learning_rate": 3.816790971997121e-05, | |
| "loss": 0.4111, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.1205673758865249, | |
| "grad_norm": 0.4474957218070915, | |
| "learning_rate": 3.8144105108078246e-05, | |
| "loss": 0.3837, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.1246200607902737, | |
| "grad_norm": 0.5410764495037073, | |
| "learning_rate": 3.81201543660571e-05, | |
| "loss": 0.3938, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.1286727456940222, | |
| "grad_norm": 0.4802006655663055, | |
| "learning_rate": 3.809605768680377e-05, | |
| "loss": 0.3854, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.132725430597771, | |
| "grad_norm": 0.4434914245555117, | |
| "learning_rate": 3.807181526438958e-05, | |
| "loss": 0.3831, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.1367781155015197, | |
| "grad_norm": 0.5209582189052376, | |
| "learning_rate": 3.8047427294059697e-05, | |
| "loss": 0.3919, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.1408308004052685, | |
| "grad_norm": 0.43234586400162894, | |
| "learning_rate": 3.802289397223145e-05, | |
| "loss": 0.3955, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.1448834853090173, | |
| "grad_norm": 0.5099183110660161, | |
| "learning_rate": 3.7998215496492854e-05, | |
| "loss": 0.3874, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.148936170212766, | |
| "grad_norm": 0.49634781654561283, | |
| "learning_rate": 3.797339206560096e-05, | |
| "loss": 0.4034, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.1529888551165146, | |
| "grad_norm": 0.4899299154682748, | |
| "learning_rate": 3.794842387948027e-05, | |
| "loss": 0.3876, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.1570415400202634, | |
| "grad_norm": 0.4328555427529725, | |
| "learning_rate": 3.7923311139221114e-05, | |
| "loss": 0.3837, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.1610942249240122, | |
| "grad_norm": 0.483664214377466, | |
| "learning_rate": 3.7898054047078054e-05, | |
| "loss": 0.3886, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.165146909827761, | |
| "grad_norm": 0.4040657262496778, | |
| "learning_rate": 3.787265280646825e-05, | |
| "loss": 0.392, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.1691995947315097, | |
| "grad_norm": 0.37025553018547264, | |
| "learning_rate": 3.7847107621969786e-05, | |
| "loss": 0.3721, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.1732522796352582, | |
| "grad_norm": 0.41879869779246093, | |
| "learning_rate": 3.7821418699320064e-05, | |
| "loss": 0.3809, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.177304964539007, | |
| "grad_norm": 0.3623034377583263, | |
| "learning_rate": 3.7795586245414145e-05, | |
| "loss": 0.3871, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.1813576494427558, | |
| "grad_norm": 0.3574498957004994, | |
| "learning_rate": 3.776961046830306e-05, | |
| "loss": 0.3833, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.1854103343465046, | |
| "grad_norm": 0.37363197224994804, | |
| "learning_rate": 3.774349157719215e-05, | |
| "loss": 0.3826, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.1894630192502533, | |
| "grad_norm": 0.3722348288190102, | |
| "learning_rate": 3.7717229782439365e-05, | |
| "loss": 0.378, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.193515704154002, | |
| "grad_norm": 0.44463920383583444, | |
| "learning_rate": 3.769082529555359e-05, | |
| "loss": 0.3993, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.1975683890577509, | |
| "grad_norm": 0.3658500163861367, | |
| "learning_rate": 3.766427832919294e-05, | |
| "loss": 0.3917, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.2016210739614994, | |
| "grad_norm": 0.4265495838126786, | |
| "learning_rate": 3.7637589097163024e-05, | |
| "loss": 0.3954, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.2056737588652482, | |
| "grad_norm": 0.33842161990918085, | |
| "learning_rate": 3.761075781441526e-05, | |
| "loss": 0.402, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.209726443768997, | |
| "grad_norm": 0.4115585421434717, | |
| "learning_rate": 3.75837846970451e-05, | |
| "loss": 0.3832, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.2137791286727457, | |
| "grad_norm": 0.42314401358331055, | |
| "learning_rate": 3.755666996229032e-05, | |
| "loss": 0.3694, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.2178318135764945, | |
| "grad_norm": 0.46271622036539484, | |
| "learning_rate": 3.752941382852927e-05, | |
| "loss": 0.381, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.221884498480243, | |
| "grad_norm": 0.43350303586921607, | |
| "learning_rate": 3.7502016515279115e-05, | |
| "loss": 0.3847, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.2259371833839918, | |
| "grad_norm": 0.37806473906491267, | |
| "learning_rate": 3.7474478243194043e-05, | |
| "loss": 0.3992, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.2299898682877406, | |
| "grad_norm": 0.36360879411492214, | |
| "learning_rate": 3.744679923406351e-05, | |
| "loss": 0.394, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.2340425531914894, | |
| "grad_norm": 0.4164246059973871, | |
| "learning_rate": 3.741897971081043e-05, | |
| "loss": 0.389, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.2380952380952381, | |
| "grad_norm": 0.3658695757856478, | |
| "learning_rate": 3.739101989748946e-05, | |
| "loss": 0.3978, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.242147922998987, | |
| "grad_norm": 0.3342649379504388, | |
| "learning_rate": 3.7362920019285066e-05, | |
| "loss": 0.3967, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.2462006079027357, | |
| "grad_norm": 0.38969919878742043, | |
| "learning_rate": 3.73346803025098e-05, | |
| "loss": 0.3847, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.2502532928064842, | |
| "grad_norm": 0.35966411286021194, | |
| "learning_rate": 3.730630097460247e-05, | |
| "loss": 0.3898, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.254305977710233, | |
| "grad_norm": 0.37205410479038303, | |
| "learning_rate": 3.727778226412628e-05, | |
| "loss": 0.3913, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.2583586626139818, | |
| "grad_norm": 0.4458058433786826, | |
| "learning_rate": 3.7249124400767006e-05, | |
| "loss": 0.3886, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.2624113475177305, | |
| "grad_norm": 0.39727715302586325, | |
| "learning_rate": 3.722032761533114e-05, | |
| "loss": 0.3988, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.2664640324214793, | |
| "grad_norm": 0.35951971001723393, | |
| "learning_rate": 3.719139213974403e-05, | |
| "loss": 0.3788, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.2705167173252279, | |
| "grad_norm": 0.42386164094553486, | |
| "learning_rate": 3.7162318207048006e-05, | |
| "loss": 0.3756, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.2745694022289766, | |
| "grad_norm": 0.40159047425911104, | |
| "learning_rate": 3.713310605140055e-05, | |
| "loss": 0.3795, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.2786220871327254, | |
| "grad_norm": 0.36750618520927125, | |
| "learning_rate": 3.710375590807233e-05, | |
| "loss": 0.3903, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.2826747720364742, | |
| "grad_norm": 0.396140389126065, | |
| "learning_rate": 3.7074268013445365e-05, | |
| "loss": 0.3655, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.286727456940223, | |
| "grad_norm": 0.41206070713763315, | |
| "learning_rate": 3.7044642605011114e-05, | |
| "loss": 0.4111, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.2907801418439715, | |
| "grad_norm": 0.42391300535126597, | |
| "learning_rate": 3.701487992136854e-05, | |
| "loss": 0.3807, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.2948328267477205, | |
| "grad_norm": 0.401484178480539, | |
| "learning_rate": 3.69849802022222e-05, | |
| "loss": 0.3921, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.298885511651469, | |
| "grad_norm": 0.4270308277004682, | |
| "learning_rate": 3.6954943688380334e-05, | |
| "loss": 0.3746, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.3029381965552178, | |
| "grad_norm": 0.40474515772402203, | |
| "learning_rate": 3.692477062175289e-05, | |
| "loss": 0.3728, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.3069908814589666, | |
| "grad_norm": 0.35749260187637605, | |
| "learning_rate": 3.689446124534958e-05, | |
| "loss": 0.3795, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.3110435663627154, | |
| "grad_norm": 0.4167189269177485, | |
| "learning_rate": 3.686401580327799e-05, | |
| "loss": 0.3874, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.3150962512664641, | |
| "grad_norm": 0.3790334172886269, | |
| "learning_rate": 3.683343454074149e-05, | |
| "loss": 0.4083, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.3191489361702127, | |
| "grad_norm": 0.3960996568712319, | |
| "learning_rate": 3.6802717704037386e-05, | |
| "loss": 0.3743, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.3232016210739614, | |
| "grad_norm": 0.3768757246665396, | |
| "learning_rate": 3.6771865540554855e-05, | |
| "loss": 0.3877, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.3272543059777102, | |
| "grad_norm": 0.49423644056149674, | |
| "learning_rate": 3.674087829877297e-05, | |
| "loss": 0.3977, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.331306990881459, | |
| "grad_norm": 0.40667255461466684, | |
| "learning_rate": 3.6709756228258735e-05, | |
| "loss": 0.3731, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.3353596757852078, | |
| "grad_norm": 0.4485553153393057, | |
| "learning_rate": 3.667849957966501e-05, | |
| "loss": 0.403, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.3394123606889563, | |
| "grad_norm": 0.4955432323976754, | |
| "learning_rate": 3.6647108604728546e-05, | |
| "loss": 0.383, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.3434650455927053, | |
| "grad_norm": 0.4051810220439692, | |
| "learning_rate": 3.661558355626795e-05, | |
| "loss": 0.3917, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.3475177304964538, | |
| "grad_norm": 0.4594629169927838, | |
| "learning_rate": 3.658392468818163e-05, | |
| "loss": 0.3717, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.3515704154002026, | |
| "grad_norm": 0.43669308855897376, | |
| "learning_rate": 3.655213225544574e-05, | |
| "loss": 0.3963, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.3556231003039514, | |
| "grad_norm": 0.49386042153598286, | |
| "learning_rate": 3.652020651411218e-05, | |
| "loss": 0.3906, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.3596757852077002, | |
| "grad_norm": 0.3913606159997072, | |
| "learning_rate": 3.6488147721306474e-05, | |
| "loss": 0.3994, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.363728470111449, | |
| "grad_norm": 0.4707759583899357, | |
| "learning_rate": 3.645595613522574e-05, | |
| "loss": 0.4007, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.3677811550151975, | |
| "grad_norm": 0.5183073784714471, | |
| "learning_rate": 3.642363201513657e-05, | |
| "loss": 0.3979, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.3718338399189463, | |
| "grad_norm": 0.36733941459468, | |
| "learning_rate": 3.6391175621373006e-05, | |
| "loss": 0.3937, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.375886524822695, | |
| "grad_norm": 0.43480500085643664, | |
| "learning_rate": 3.6358587215334355e-05, | |
| "loss": 0.3958, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.3799392097264438, | |
| "grad_norm": 0.45301319787141103, | |
| "learning_rate": 3.632586705948318e-05, | |
| "loss": 0.3929, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.3839918946301926, | |
| "grad_norm": 0.41020751739720523, | |
| "learning_rate": 3.629301541734311e-05, | |
| "loss": 0.3738, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.3880445795339411, | |
| "grad_norm": 0.49566671479533825, | |
| "learning_rate": 3.626003255349676e-05, | |
| "loss": 0.3857, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.39209726443769, | |
| "grad_norm": 0.45144467086471224, | |
| "learning_rate": 3.622691873358357e-05, | |
| "loss": 0.4101, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.3961499493414387, | |
| "grad_norm": 0.418247196909798, | |
| "learning_rate": 3.61936742242977e-05, | |
| "loss": 0.4074, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.4002026342451874, | |
| "grad_norm": 0.3927422779585475, | |
| "learning_rate": 3.6160299293385864e-05, | |
| "loss": 0.406, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.4042553191489362, | |
| "grad_norm": 0.37287093304492575, | |
| "learning_rate": 3.612679420964516e-05, | |
| "loss": 0.4078, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.408308004052685, | |
| "grad_norm": 0.4240051117770986, | |
| "learning_rate": 3.609315924292092e-05, | |
| "loss": 0.3825, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.4123606889564337, | |
| "grad_norm": 0.4408499075682911, | |
| "learning_rate": 3.6059394664104554e-05, | |
| "loss": 0.3966, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.4164133738601823, | |
| "grad_norm": 0.458480615701739, | |
| "learning_rate": 3.602550074513133e-05, | |
| "loss": 0.3894, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.420466058763931, | |
| "grad_norm": 0.4165066202735225, | |
| "learning_rate": 3.599147775897822e-05, | |
| "loss": 0.3783, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.4245187436676798, | |
| "grad_norm": 0.3461737782259951, | |
| "learning_rate": 3.595732597966167e-05, | |
| "loss": 0.3893, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.4285714285714286, | |
| "grad_norm": 0.4065034862171919, | |
| "learning_rate": 3.592304568223542e-05, | |
| "loss": 0.3945, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.4326241134751774, | |
| "grad_norm": 0.40485465672951415, | |
| "learning_rate": 3.588863714278826e-05, | |
| "loss": 0.3565, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.436676798378926, | |
| "grad_norm": 0.4928460824588013, | |
| "learning_rate": 3.585410063844186e-05, | |
| "loss": 0.3902, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.4407294832826747, | |
| "grad_norm": 0.4769368241015664, | |
| "learning_rate": 3.581943644734846e-05, | |
| "loss": 0.4056, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.4447821681864235, | |
| "grad_norm": 0.4518968783825401, | |
| "learning_rate": 3.578464484868869e-05, | |
| "loss": 0.3778, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.4488348530901722, | |
| "grad_norm": 0.44986120014669656, | |
| "learning_rate": 3.5749726122669316e-05, | |
| "loss": 0.397, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.452887537993921, | |
| "grad_norm": 0.3658325292439502, | |
| "learning_rate": 3.5714680550520943e-05, | |
| "loss": 0.3954, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.4569402228976698, | |
| "grad_norm": 0.40690602802032383, | |
| "learning_rate": 3.5679508414495794e-05, | |
| "loss": 0.381, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.4609929078014185, | |
| "grad_norm": 0.31311610432042475, | |
| "learning_rate": 3.564420999786543e-05, | |
| "loss": 0.3959, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.465045592705167, | |
| "grad_norm": 0.46129267162685317, | |
| "learning_rate": 3.560878558491842e-05, | |
| "loss": 0.3911, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.4690982776089159, | |
| "grad_norm": 0.4048412644955191, | |
| "learning_rate": 3.5573235460958145e-05, | |
| "loss": 0.3863, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.4731509625126646, | |
| "grad_norm": 0.3309528950813826, | |
| "learning_rate": 3.553755991230039e-05, | |
| "loss": 0.3771, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.4772036474164134, | |
| "grad_norm": 0.404255918455759, | |
| "learning_rate": 3.5501759226271144e-05, | |
| "loss": 0.386, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.4812563323201622, | |
| "grad_norm": 0.32528191716305815, | |
| "learning_rate": 3.546583369120419e-05, | |
| "loss": 0.3862, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.4853090172239107, | |
| "grad_norm": 0.3893528575359535, | |
| "learning_rate": 3.5429783596438864e-05, | |
| "loss": 0.398, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.4893617021276595, | |
| "grad_norm": 0.36126126749891163, | |
| "learning_rate": 3.539360923231766e-05, | |
| "loss": 0.39, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.4934143870314083, | |
| "grad_norm": 0.32383796343790616, | |
| "learning_rate": 3.535731089018394e-05, | |
| "loss": 0.3797, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.497467071935157, | |
| "grad_norm": 0.4070970502250912, | |
| "learning_rate": 3.532088886237956e-05, | |
| "loss": 0.3789, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.5015197568389058, | |
| "grad_norm": 0.36710692301197073, | |
| "learning_rate": 3.528434344224253e-05, | |
| "loss": 0.3854, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.5055724417426544, | |
| "grad_norm": 0.3198642111518589, | |
| "learning_rate": 3.524767492410464e-05, | |
| "loss": 0.3953, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.5096251266464034, | |
| "grad_norm": 0.31607338002139157, | |
| "learning_rate": 3.521088360328908e-05, | |
| "loss": 0.3726, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.513677811550152, | |
| "grad_norm": 0.35580429448891365, | |
| "learning_rate": 3.517396977610811e-05, | |
| "loss": 0.3813, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.5177304964539007, | |
| "grad_norm": 0.3220053924852848, | |
| "learning_rate": 3.5136933739860595e-05, | |
| "loss": 0.4056, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.5217831813576495, | |
| "grad_norm": 0.3343340545339712, | |
| "learning_rate": 3.509977579282971e-05, | |
| "loss": 0.3957, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.5258358662613982, | |
| "grad_norm": 0.40392176265054086, | |
| "learning_rate": 3.5062496234280424e-05, | |
| "loss": 0.3992, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.529888551165147, | |
| "grad_norm": 0.33579673639540947, | |
| "learning_rate": 3.502509536445719e-05, | |
| "loss": 0.3766, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.5339412360688955, | |
| "grad_norm": 0.3244782685872029, | |
| "learning_rate": 3.498757348458147e-05, | |
| "loss": 0.3668, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.5379939209726445, | |
| "grad_norm": 0.35842410026300425, | |
| "learning_rate": 3.4949930896849324e-05, | |
| "loss": 0.3755, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.542046605876393, | |
| "grad_norm": 0.38293450673581925, | |
| "learning_rate": 3.491216790442899e-05, | |
| "loss": 0.3761, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.5460992907801419, | |
| "grad_norm": 0.3305168953142244, | |
| "learning_rate": 3.487428481145839e-05, | |
| "loss": 0.4018, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.5501519756838906, | |
| "grad_norm": 0.3844280444720269, | |
| "learning_rate": 3.483628192304278e-05, | |
| "loss": 0.3918, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.5542046605876392, | |
| "grad_norm": 0.3486126278871181, | |
| "learning_rate": 3.479815954525219e-05, | |
| "loss": 0.38, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.5582573454913882, | |
| "grad_norm": 0.3481002460119227, | |
| "learning_rate": 3.475991798511899e-05, | |
| "loss": 0.3809, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.5623100303951367, | |
| "grad_norm": 0.3434419528883373, | |
| "learning_rate": 3.4721557550635464e-05, | |
| "loss": 0.3911, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.5663627152988855, | |
| "grad_norm": 0.3541905245245948, | |
| "learning_rate": 3.468307855075128e-05, | |
| "loss": 0.3843, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.5704154002026343, | |
| "grad_norm": 0.3528131640034027, | |
| "learning_rate": 3.4644481295371005e-05, | |
| "loss": 0.3802, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.574468085106383, | |
| "grad_norm": 0.32248705956101925, | |
| "learning_rate": 3.460576609535163e-05, | |
| "loss": 0.3817, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.5785207700101318, | |
| "grad_norm": 0.38736052892826134, | |
| "learning_rate": 3.456693326250006e-05, | |
| "loss": 0.3895, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.5825734549138804, | |
| "grad_norm": 0.29970674710617945, | |
| "learning_rate": 3.452798310957058e-05, | |
| "loss": 0.388, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.5866261398176293, | |
| "grad_norm": 0.35382269712288744, | |
| "learning_rate": 3.4488915950262386e-05, | |
| "loss": 0.3807, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.590678824721378, | |
| "grad_norm": 0.37313223931684336, | |
| "learning_rate": 3.4449732099216985e-05, | |
| "loss": 0.3798, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.5947315096251267, | |
| "grad_norm": 0.3564085108524849, | |
| "learning_rate": 3.441043187201574e-05, | |
| "loss": 0.3873, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.5987841945288754, | |
| "grad_norm": 0.3896744705660433, | |
| "learning_rate": 3.437101558517728e-05, | |
| "loss": 0.379, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.602836879432624, | |
| "grad_norm": 0.4236477269790274, | |
| "learning_rate": 3.433148355615496e-05, | |
| "loss": 0.3805, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.606889564336373, | |
| "grad_norm": 0.33180303487846063, | |
| "learning_rate": 3.4291836103334294e-05, | |
| "loss": 0.3816, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.6109422492401215, | |
| "grad_norm": 0.40046298740007275, | |
| "learning_rate": 3.425207354603043e-05, | |
| "loss": 0.3919, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.6149949341438703, | |
| "grad_norm": 0.328353435367082, | |
| "learning_rate": 3.421219620448553e-05, | |
| "loss": 0.372, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.619047619047619, | |
| "grad_norm": 0.41157638880916503, | |
| "learning_rate": 3.417220439986623e-05, | |
| "loss": 0.3823, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.6231003039513676, | |
| "grad_norm": 0.34310808322575453, | |
| "learning_rate": 3.4132098454261024e-05, | |
| "loss": 0.3912, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.6271529888551166, | |
| "grad_norm": 0.351897654629441, | |
| "learning_rate": 3.4091878690677676e-05, | |
| "loss": 0.366, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.6312056737588652, | |
| "grad_norm": 0.4088255638985235, | |
| "learning_rate": 3.405154543304065e-05, | |
| "loss": 0.3827, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.635258358662614, | |
| "grad_norm": 0.33148711590680646, | |
| "learning_rate": 3.401109900618843e-05, | |
| "loss": 0.3906, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.6393110435663627, | |
| "grad_norm": 0.34883575253586685, | |
| "learning_rate": 3.3970539735870996e-05, | |
| "loss": 0.3898, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.6433637284701115, | |
| "grad_norm": 0.3332468043300822, | |
| "learning_rate": 3.392986794874714e-05, | |
| "loss": 0.3894, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.6474164133738602, | |
| "grad_norm": 0.38444511357590755, | |
| "learning_rate": 3.388908397238184e-05, | |
| "loss": 0.3876, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.6514690982776088, | |
| "grad_norm": 0.33709272088240394, | |
| "learning_rate": 3.384818813524362e-05, | |
| "loss": 0.3683, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.6555217831813578, | |
| "grad_norm": 0.41679613998216886, | |
| "learning_rate": 3.380718076670195e-05, | |
| "loss": 0.3725, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.6595744680851063, | |
| "grad_norm": 0.33398358453995164, | |
| "learning_rate": 3.376606219702454e-05, | |
| "loss": 0.3852, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.663627152988855, | |
| "grad_norm": 0.40380598044070437, | |
| "learning_rate": 3.372483275737468e-05, | |
| "loss": 0.3827, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.6676798378926039, | |
| "grad_norm": 0.37253370186815443, | |
| "learning_rate": 3.368349277980861e-05, | |
| "loss": 0.3829, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.6717325227963524, | |
| "grad_norm": 0.31919959473989934, | |
| "learning_rate": 3.3642042597272844e-05, | |
| "loss": 0.3718, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.6757852077001014, | |
| "grad_norm": 0.33904759027302445, | |
| "learning_rate": 3.360048254360144e-05, | |
| "loss": 0.3839, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.67983789260385, | |
| "grad_norm": 0.3026600003801669, | |
| "learning_rate": 3.355881295351336e-05, | |
| "loss": 0.3852, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.6838905775075987, | |
| "grad_norm": 0.37635097905957826, | |
| "learning_rate": 3.351703416260975e-05, | |
| "loss": 0.3963, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.6879432624113475, | |
| "grad_norm": 0.32908616236192817, | |
| "learning_rate": 3.347514650737126e-05, | |
| "loss": 0.3873, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.6919959473150963, | |
| "grad_norm": 0.3480715206597894, | |
| "learning_rate": 3.3433150325155295e-05, | |
| "loss": 0.3784, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.696048632218845, | |
| "grad_norm": 0.40580832061545785, | |
| "learning_rate": 3.339104595419334e-05, | |
| "loss": 0.3818, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.7001013171225936, | |
| "grad_norm": 0.3448601335012333, | |
| "learning_rate": 3.3348833733588204e-05, | |
| "loss": 0.3798, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.7041540020263426, | |
| "grad_norm": 0.3872028822842741, | |
| "learning_rate": 3.3306514003311305e-05, | |
| "loss": 0.3778, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.7082066869300911, | |
| "grad_norm": 0.3183836838920985, | |
| "learning_rate": 3.326408710419996e-05, | |
| "loss": 0.3937, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.71225937183384, | |
| "grad_norm": 0.3748241260527875, | |
| "learning_rate": 3.322155337795454e-05, | |
| "loss": 0.3824, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.7163120567375887, | |
| "grad_norm": 0.404732639180344, | |
| "learning_rate": 3.317891316713587e-05, | |
| "loss": 0.3926, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.7203647416413372, | |
| "grad_norm": 0.38031442212867883, | |
| "learning_rate": 3.313616681516231e-05, | |
| "loss": 0.3771, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.7244174265450862, | |
| "grad_norm": 0.35108666496500895, | |
| "learning_rate": 3.309331466630713e-05, | |
| "loss": 0.3861, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.7284701114488348, | |
| "grad_norm": 0.3996339579834394, | |
| "learning_rate": 3.305035706569563e-05, | |
| "loss": 0.3721, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.7325227963525835, | |
| "grad_norm": 0.391047444366928, | |
| "learning_rate": 3.3007294359302433e-05, | |
| "loss": 0.3948, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.7365754812563323, | |
| "grad_norm": 0.3930790482552998, | |
| "learning_rate": 3.296412689394864e-05, | |
| "loss": 0.3883, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.740628166160081, | |
| "grad_norm": 0.5009590621870026, | |
| "learning_rate": 3.292085501729909e-05, | |
| "loss": 0.3846, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.7446808510638299, | |
| "grad_norm": 0.3495775404343688, | |
| "learning_rate": 3.2877479077859534e-05, | |
| "loss": 0.3841, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.7487335359675784, | |
| "grad_norm": 0.45529883100544044, | |
| "learning_rate": 3.283399942497381e-05, | |
| "loss": 0.3759, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.7527862208713274, | |
| "grad_norm": 0.43616723066782304, | |
| "learning_rate": 3.279041640882108e-05, | |
| "loss": 0.3716, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.756838905775076, | |
| "grad_norm": 0.3475437577220352, | |
| "learning_rate": 3.2746730380412964e-05, | |
| "loss": 0.3755, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.7608915906788247, | |
| "grad_norm": 0.33976864991647565, | |
| "learning_rate": 3.2702941691590726e-05, | |
| "loss": 0.3778, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.7649442755825735, | |
| "grad_norm": 0.34752789661571815, | |
| "learning_rate": 3.265905069502244e-05, | |
| "loss": 0.3857, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.768996960486322, | |
| "grad_norm": 0.3402655086419584, | |
| "learning_rate": 3.261505774420016e-05, | |
| "loss": 0.3813, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.773049645390071, | |
| "grad_norm": 0.3165275078026653, | |
| "learning_rate": 3.257096319343707e-05, | |
| "loss": 0.3855, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.7771023302938196, | |
| "grad_norm": 0.3368800373214777, | |
| "learning_rate": 3.2526767397864614e-05, | |
| "loss": 0.3816, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.7811550151975684, | |
| "grad_norm": 0.30624347633869614, | |
| "learning_rate": 3.248247071342966e-05, | |
| "loss": 0.3947, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.7852077001013171, | |
| "grad_norm": 0.32837568594558086, | |
| "learning_rate": 3.243807349689161e-05, | |
| "loss": 0.382, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.789260385005066, | |
| "grad_norm": 0.3321459765634274, | |
| "learning_rate": 3.2393576105819544e-05, | |
| "loss": 0.3868, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.7933130699088147, | |
| "grad_norm": 0.31400864852152416, | |
| "learning_rate": 3.2348978898589333e-05, | |
| "loss": 0.3865, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.7973657548125632, | |
| "grad_norm": 0.38648048855113326, | |
| "learning_rate": 3.230428223438075e-05, | |
| "loss": 0.3975, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.8014184397163122, | |
| "grad_norm": 0.3644702704091603, | |
| "learning_rate": 3.225948647317459e-05, | |
| "loss": 0.3775, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.8054711246200608, | |
| "grad_norm": 0.30465682616597806, | |
| "learning_rate": 3.2214591975749745e-05, | |
| "loss": 0.3964, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.8095238095238095, | |
| "grad_norm": 0.38912777011083466, | |
| "learning_rate": 3.216959910368034e-05, | |
| "loss": 0.3849, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.8135764944275583, | |
| "grad_norm": 0.3742440014325111, | |
| "learning_rate": 3.212450821933277e-05, | |
| "loss": 0.397, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.8176291793313069, | |
| "grad_norm": 0.3198043525293136, | |
| "learning_rate": 3.207931968586281e-05, | |
| "loss": 0.394, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.8216818642350558, | |
| "grad_norm": 0.40581679630478645, | |
| "learning_rate": 3.203403386721272e-05, | |
| "loss": 0.3752, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.8257345491388044, | |
| "grad_norm": 0.28706617378651644, | |
| "learning_rate": 3.1988651128108245e-05, | |
| "loss": 0.3806, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.8297872340425532, | |
| "grad_norm": 0.38879030547963733, | |
| "learning_rate": 3.194317183405573e-05, | |
| "loss": 0.3837, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.833839918946302, | |
| "grad_norm": 0.3421366570186139, | |
| "learning_rate": 3.189759635133914e-05, | |
| "loss": 0.3936, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.8378926038500507, | |
| "grad_norm": 0.37032058180455646, | |
| "learning_rate": 3.185192504701718e-05, | |
| "loss": 0.3829, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.8419452887537995, | |
| "grad_norm": 0.4098360610185781, | |
| "learning_rate": 3.1806158288920234e-05, | |
| "loss": 0.3871, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.845997973657548, | |
| "grad_norm": 0.31274440927101876, | |
| "learning_rate": 3.1760296445647477e-05, | |
| "loss": 0.374, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.850050658561297, | |
| "grad_norm": 0.3746035718310915, | |
| "learning_rate": 3.1714339886563896e-05, | |
| "loss": 0.3856, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.8541033434650456, | |
| "grad_norm": 0.33482962338135475, | |
| "learning_rate": 3.166828898179731e-05, | |
| "loss": 0.373, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.8581560283687943, | |
| "grad_norm": 0.33558439918222477, | |
| "learning_rate": 3.162214410223536e-05, | |
| "loss": 0.3807, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.8622087132725431, | |
| "grad_norm": 0.41895181904577067, | |
| "learning_rate": 3.157590561952257e-05, | |
| "loss": 0.3833, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.8662613981762917, | |
| "grad_norm": 0.3453305267625051, | |
| "learning_rate": 3.152957390605732e-05, | |
| "loss": 0.3893, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.8703140830800407, | |
| "grad_norm": 0.3453542013732836, | |
| "learning_rate": 3.148314933498886e-05, | |
| "loss": 0.3859, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.8743667679837892, | |
| "grad_norm": 0.42577012305472484, | |
| "learning_rate": 3.143663228021431e-05, | |
| "loss": 0.3938, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.878419452887538, | |
| "grad_norm": 0.33177522473893084, | |
| "learning_rate": 3.1390023116375624e-05, | |
| "loss": 0.3846, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.8824721377912867, | |
| "grad_norm": 0.35691192252799064, | |
| "learning_rate": 3.134332221885661e-05, | |
| "loss": 0.3812, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.8865248226950353, | |
| "grad_norm": 0.3717780154365357, | |
| "learning_rate": 3.129652996377987e-05, | |
| "loss": 0.389, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.8905775075987843, | |
| "grad_norm": 0.2878577760310673, | |
| "learning_rate": 3.12496467280038e-05, | |
| "loss": 0.3837, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.8946301925025328, | |
| "grad_norm": 0.35828864412395905, | |
| "learning_rate": 3.120267288911952e-05, | |
| "loss": 0.3771, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.8986828774062816, | |
| "grad_norm": 0.3008387494480531, | |
| "learning_rate": 3.11556088254479e-05, | |
| "loss": 0.3794, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.9027355623100304, | |
| "grad_norm": 0.3293872231335726, | |
| "learning_rate": 3.11084549160364e-05, | |
| "loss": 0.3758, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.9067882472137792, | |
| "grad_norm": 0.3208863798171764, | |
| "learning_rate": 3.106121154065615e-05, | |
| "loss": 0.3912, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.910840932117528, | |
| "grad_norm": 0.3038992052076309, | |
| "learning_rate": 3.1013879079798805e-05, | |
| "loss": 0.3855, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.9148936170212765, | |
| "grad_norm": 0.30942001788690743, | |
| "learning_rate": 3.096645791467348e-05, | |
| "loss": 0.3843, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.9189463019250255, | |
| "grad_norm": 0.32682436248549696, | |
| "learning_rate": 3.091894842720373e-05, | |
| "loss": 0.3807, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.922998986828774, | |
| "grad_norm": 0.28421862405370174, | |
| "learning_rate": 3.0871351000024425e-05, | |
| "loss": 0.3862, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.9270516717325228, | |
| "grad_norm": 0.31454189950997585, | |
| "learning_rate": 3.0823666016478716e-05, | |
| "loss": 0.3755, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.9311043566362716, | |
| "grad_norm": 0.3076517474348167, | |
| "learning_rate": 3.0775893860614896e-05, | |
| "loss": 0.3747, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.93515704154002, | |
| "grad_norm": 0.337462676255627, | |
| "learning_rate": 3.0728034917183336e-05, | |
| "loss": 0.3864, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.939209726443769, | |
| "grad_norm": 0.32952554709262866, | |
| "learning_rate": 3.06800895716334e-05, | |
| "loss": 0.3972, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.9432624113475176, | |
| "grad_norm": 0.3082096124108131, | |
| "learning_rate": 3.063205821011029e-05, | |
| "loss": 0.3736, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.9473150962512664, | |
| "grad_norm": 0.32385988766280205, | |
| "learning_rate": 3.0583941219452016e-05, | |
| "loss": 0.3691, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.9513677811550152, | |
| "grad_norm": 0.2806282495159555, | |
| "learning_rate": 3.053573898718618e-05, | |
| "loss": 0.3946, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.955420466058764, | |
| "grad_norm": 0.3955481253736078, | |
| "learning_rate": 3.0487451901526956e-05, | |
| "loss": 0.3718, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.9594731509625127, | |
| "grad_norm": 0.3114353744119773, | |
| "learning_rate": 3.0439080351371875e-05, | |
| "loss": 0.3797, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.9635258358662613, | |
| "grad_norm": 0.34789856413421266, | |
| "learning_rate": 3.0390624726298764e-05, | |
| "loss": 0.3884, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.9675785207700103, | |
| "grad_norm": 0.30212523956271187, | |
| "learning_rate": 3.034208541656255e-05, | |
| "loss": 0.3985, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.9716312056737588, | |
| "grad_norm": 0.2875510907276814, | |
| "learning_rate": 3.029346281309218e-05, | |
| "loss": 0.3857, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.9756838905775076, | |
| "grad_norm": 0.3135530034694446, | |
| "learning_rate": 3.0244757307487415e-05, | |
| "loss": 0.3819, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.9797365754812564, | |
| "grad_norm": 0.274428402960773, | |
| "learning_rate": 3.019596929201569e-05, | |
| "loss": 0.3763, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.983789260385005, | |
| "grad_norm": 0.29962109864446534, | |
| "learning_rate": 3.0147099159608985e-05, | |
| "loss": 0.3794, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.987841945288754, | |
| "grad_norm": 0.2876375055781074, | |
| "learning_rate": 3.0098147303860616e-05, | |
| "loss": 0.3868, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.9918946301925025, | |
| "grad_norm": 0.2807904680045187, | |
| "learning_rate": 3.0049114119022117e-05, | |
| "loss": 0.3856, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.9959473150962512, | |
| "grad_norm": 0.3079166326958009, | |
| "learning_rate": 3.0000000000000004e-05, | |
| "loss": 0.3936, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 2.0020263424518743, | |
| "grad_norm": 0.33927022459155104, | |
| "learning_rate": 2.995080534235264e-05, | |
| "loss": 0.3766, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 2.0060790273556233, | |
| "grad_norm": 0.455854065164269, | |
| "learning_rate": 2.9901530542287044e-05, | |
| "loss": 0.3212, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 2.010131712259372, | |
| "grad_norm": 1.242600810595643, | |
| "learning_rate": 2.9852175996655676e-05, | |
| "loss": 0.3317, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.0141843971631204, | |
| "grad_norm": 0.6332210645632647, | |
| "learning_rate": 2.980274210295326e-05, | |
| "loss": 0.3276, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 2.0182370820668694, | |
| "grad_norm": 0.4946693260016561, | |
| "learning_rate": 2.9753229259313578e-05, | |
| "loss": 0.3202, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 2.022289766970618, | |
| "grad_norm": 0.4683996859560435, | |
| "learning_rate": 2.9703637864506274e-05, | |
| "loss": 0.3083, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 2.026342451874367, | |
| "grad_norm": 0.5664790595418052, | |
| "learning_rate": 2.965396831793362e-05, | |
| "loss": 0.3147, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 2.0303951367781155, | |
| "grad_norm": 0.6345057937361971, | |
| "learning_rate": 2.9604221019627316e-05, | |
| "loss": 0.3203, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0344478216818644, | |
| "grad_norm": 0.415262731072021, | |
| "learning_rate": 2.955439637024526e-05, | |
| "loss": 0.3232, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 2.038500506585613, | |
| "grad_norm": 0.6033620823978579, | |
| "learning_rate": 2.9504494771068334e-05, | |
| "loss": 0.3248, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 2.0425531914893615, | |
| "grad_norm": 0.5088836020263809, | |
| "learning_rate": 2.9454516623997156e-05, | |
| "loss": 0.3022, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 2.0466058763931105, | |
| "grad_norm": 0.5414167845397915, | |
| "learning_rate": 2.9404462331548847e-05, | |
| "loss": 0.3181, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 2.050658561296859, | |
| "grad_norm": 0.6682648280437532, | |
| "learning_rate": 2.93543322968538e-05, | |
| "loss": 0.3149, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 2.054711246200608, | |
| "grad_norm": 0.36811811814719536, | |
| "learning_rate": 2.9304126923652428e-05, | |
| "loss": 0.2998, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 2.0587639311043566, | |
| "grad_norm": 0.6550259300279986, | |
| "learning_rate": 2.9253846616291896e-05, | |
| "loss": 0.32, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 2.062816616008105, | |
| "grad_norm": 0.41987616897819174, | |
| "learning_rate": 2.9203491779722896e-05, | |
| "loss": 0.3319, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 2.066869300911854, | |
| "grad_norm": 0.5205799167755208, | |
| "learning_rate": 2.9153062819496357e-05, | |
| "loss": 0.3138, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 2.0709219858156027, | |
| "grad_norm": 0.41107693360288877, | |
| "learning_rate": 2.9102560141760178e-05, | |
| "loss": 0.3231, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.0749746707193517, | |
| "grad_norm": 0.42879617365550077, | |
| "learning_rate": 2.9051984153256004e-05, | |
| "loss": 0.3158, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 2.0790273556231003, | |
| "grad_norm": 0.42454692594093485, | |
| "learning_rate": 2.900133526131588e-05, | |
| "loss": 0.3163, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 2.0830800405268493, | |
| "grad_norm": 0.3657562373074371, | |
| "learning_rate": 2.8950613873859025e-05, | |
| "loss": 0.3174, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 2.087132725430598, | |
| "grad_norm": 0.3436010709070961, | |
| "learning_rate": 2.8899820399388515e-05, | |
| "loss": 0.3027, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 2.0911854103343464, | |
| "grad_norm": 0.34051541825058873, | |
| "learning_rate": 2.8848955246988012e-05, | |
| "loss": 0.3183, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 2.0952380952380953, | |
| "grad_norm": 0.3120749866187744, | |
| "learning_rate": 2.879801882631847e-05, | |
| "loss": 0.3282, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 2.099290780141844, | |
| "grad_norm": 0.39674746207201744, | |
| "learning_rate": 2.8747011547614808e-05, | |
| "loss": 0.2908, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 2.103343465045593, | |
| "grad_norm": 0.2782864066683504, | |
| "learning_rate": 2.8695933821682635e-05, | |
| "loss": 0.3024, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 2.1073961499493414, | |
| "grad_norm": 0.4167458341397558, | |
| "learning_rate": 2.864478605989494e-05, | |
| "loss": 0.3203, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 2.11144883485309, | |
| "grad_norm": 0.359870235105583, | |
| "learning_rate": 2.8593568674188765e-05, | |
| "loss": 0.3267, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.115501519756839, | |
| "grad_norm": 0.3638860753368854, | |
| "learning_rate": 2.8542282077061892e-05, | |
| "loss": 0.3148, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 2.1195542046605875, | |
| "grad_norm": 0.3876326476810913, | |
| "learning_rate": 2.8490926681569523e-05, | |
| "loss": 0.3084, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 2.1236068895643365, | |
| "grad_norm": 0.3545809967558409, | |
| "learning_rate": 2.8439502901320956e-05, | |
| "loss": 0.3009, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 2.127659574468085, | |
| "grad_norm": 0.38300385979221807, | |
| "learning_rate": 2.8388011150476237e-05, | |
| "loss": 0.3235, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 2.1317122593718336, | |
| "grad_norm": 0.3396183283945086, | |
| "learning_rate": 2.8336451843742866e-05, | |
| "loss": 0.3117, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.1357649442755826, | |
| "grad_norm": 0.31612539768425013, | |
| "learning_rate": 2.8284825396372387e-05, | |
| "loss": 0.3158, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 2.139817629179331, | |
| "grad_norm": 0.2936800900590169, | |
| "learning_rate": 2.8233132224157132e-05, | |
| "loss": 0.3137, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 2.14387031408308, | |
| "grad_norm": 0.3000615237264944, | |
| "learning_rate": 2.8181372743426805e-05, | |
| "loss": 0.3246, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 2.1479229989868287, | |
| "grad_norm": 0.29688365756112967, | |
| "learning_rate": 2.8129547371045128e-05, | |
| "loss": 0.3172, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 2.1519756838905777, | |
| "grad_norm": 0.2942073478084988, | |
| "learning_rate": 2.8077656524406534e-05, | |
| "loss": 0.3095, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.1560283687943262, | |
| "grad_norm": 0.3216458481400994, | |
| "learning_rate": 2.802570062143278e-05, | |
| "loss": 0.3218, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 2.160081053698075, | |
| "grad_norm": 0.3146567128293041, | |
| "learning_rate": 2.7973680080569555e-05, | |
| "loss": 0.3037, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 2.164133738601824, | |
| "grad_norm": 0.32742738457895465, | |
| "learning_rate": 2.792159532078314e-05, | |
| "loss": 0.3298, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 2.1681864235055723, | |
| "grad_norm": 0.3064838094697248, | |
| "learning_rate": 2.7869446761557033e-05, | |
| "loss": 0.2986, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 2.1722391084093213, | |
| "grad_norm": 0.27817813040469985, | |
| "learning_rate": 2.781723482288857e-05, | |
| "loss": 0.3106, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 2.17629179331307, | |
| "grad_norm": 0.2657930165096628, | |
| "learning_rate": 2.7764959925285517e-05, | |
| "loss": 0.3109, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 2.180344478216819, | |
| "grad_norm": 0.2811781500263199, | |
| "learning_rate": 2.771262248976272e-05, | |
| "loss": 0.3214, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 2.1843971631205674, | |
| "grad_norm": 0.2910393963763764, | |
| "learning_rate": 2.7660222937838677e-05, | |
| "loss": 0.3361, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 2.188449848024316, | |
| "grad_norm": 0.2766840764562337, | |
| "learning_rate": 2.7607761691532186e-05, | |
| "loss": 0.314, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 2.192502532928065, | |
| "grad_norm": 0.30696831726635976, | |
| "learning_rate": 2.7555239173358916e-05, | |
| "loss": 0.3078, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.1965552178318135, | |
| "grad_norm": 0.297299022486252, | |
| "learning_rate": 2.7502655806328e-05, | |
| "loss": 0.3173, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 2.2006079027355625, | |
| "grad_norm": 0.2929676957377111, | |
| "learning_rate": 2.7450012013938648e-05, | |
| "loss": 0.319, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 2.204660587639311, | |
| "grad_norm": 0.3123621083485315, | |
| "learning_rate": 2.739730822017673e-05, | |
| "loss": 0.3158, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 2.2087132725430596, | |
| "grad_norm": 0.32065599665688804, | |
| "learning_rate": 2.7344544849511355e-05, | |
| "loss": 0.3238, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 2.2127659574468086, | |
| "grad_norm": 0.2807524490357412, | |
| "learning_rate": 2.7291722326891456e-05, | |
| "loss": 0.3062, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 2.216818642350557, | |
| "grad_norm": 0.29859082286910454, | |
| "learning_rate": 2.723884107774236e-05, | |
| "loss": 0.3032, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 2.220871327254306, | |
| "grad_norm": 0.28021271513463575, | |
| "learning_rate": 2.718590152796239e-05, | |
| "loss": 0.3097, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 2.2249240121580547, | |
| "grad_norm": 0.33752544573817456, | |
| "learning_rate": 2.71329041039194e-05, | |
| "loss": 0.3182, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 2.2289766970618032, | |
| "grad_norm": 0.31079722298856244, | |
| "learning_rate": 2.7079849232447357e-05, | |
| "loss": 0.3137, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 2.2330293819655522, | |
| "grad_norm": 0.3170587668632497, | |
| "learning_rate": 2.7026737340842895e-05, | |
| "loss": 0.3083, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.237082066869301, | |
| "grad_norm": 0.2852300199291198, | |
| "learning_rate": 2.697356885686189e-05, | |
| "loss": 0.3085, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 2.2411347517730498, | |
| "grad_norm": 0.30679949164736214, | |
| "learning_rate": 2.6920344208716014e-05, | |
| "loss": 0.323, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 2.2451874366767983, | |
| "grad_norm": 0.3047374500723771, | |
| "learning_rate": 2.6867063825069252e-05, | |
| "loss": 0.3196, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 2.2492401215805473, | |
| "grad_norm": 0.3418700654001792, | |
| "learning_rate": 2.6813728135034494e-05, | |
| "loss": 0.3136, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 2.253292806484296, | |
| "grad_norm": 0.30945446267993043, | |
| "learning_rate": 2.6760337568170056e-05, | |
| "loss": 0.3049, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 2.2573454913880444, | |
| "grad_norm": 0.30840811853413186, | |
| "learning_rate": 2.6706892554476226e-05, | |
| "loss": 0.3266, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 2.2613981762917934, | |
| "grad_norm": 0.3148792056971951, | |
| "learning_rate": 2.6653393524391795e-05, | |
| "loss": 0.3137, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 2.265450861195542, | |
| "grad_norm": 0.28409939282494145, | |
| "learning_rate": 2.6599840908790592e-05, | |
| "loss": 0.3162, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 2.269503546099291, | |
| "grad_norm": 0.3319331846472534, | |
| "learning_rate": 2.6546235138978028e-05, | |
| "loss": 0.3275, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 2.2735562310030395, | |
| "grad_norm": 0.3186424611034407, | |
| "learning_rate": 2.6492576646687597e-05, | |
| "loss": 0.3257, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.2776089159067885, | |
| "grad_norm": 0.3073639901654007, | |
| "learning_rate": 2.6438865864077425e-05, | |
| "loss": 0.3344, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 2.281661600810537, | |
| "grad_norm": 0.2888297327773877, | |
| "learning_rate": 2.6385103223726766e-05, | |
| "loss": 0.3136, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 2.2857142857142856, | |
| "grad_norm": 0.3230272348724275, | |
| "learning_rate": 2.6331289158632537e-05, | |
| "loss": 0.3371, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 2.2897669706180346, | |
| "grad_norm": 0.3459693075166934, | |
| "learning_rate": 2.6277424102205817e-05, | |
| "loss": 0.3192, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 2.293819655521783, | |
| "grad_norm": 0.28762752422654336, | |
| "learning_rate": 2.6223508488268374e-05, | |
| "loss": 0.3027, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.297872340425532, | |
| "grad_norm": 0.35587379379824174, | |
| "learning_rate": 2.6169542751049148e-05, | |
| "loss": 0.3159, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 2.3019250253292807, | |
| "grad_norm": 0.3072332000790235, | |
| "learning_rate": 2.6115527325180754e-05, | |
| "loss": 0.3118, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 2.305977710233029, | |
| "grad_norm": 0.2899781064053684, | |
| "learning_rate": 2.606146264569603e-05, | |
| "loss": 0.3235, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 2.310030395136778, | |
| "grad_norm": 0.34198123380155315, | |
| "learning_rate": 2.6007349148024447e-05, | |
| "loss": 0.3102, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 2.3140830800405268, | |
| "grad_norm": 0.2708933533596076, | |
| "learning_rate": 2.5953187267988694e-05, | |
| "loss": 0.3117, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.3181357649442758, | |
| "grad_norm": 0.3261583355793608, | |
| "learning_rate": 2.5898977441801097e-05, | |
| "loss": 0.321, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 2.3221884498480243, | |
| "grad_norm": 0.26187475078449984, | |
| "learning_rate": 2.584472010606015e-05, | |
| "loss": 0.3225, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 2.326241134751773, | |
| "grad_norm": 0.33132199024455566, | |
| "learning_rate": 2.5790415697746976e-05, | |
| "loss": 0.3198, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 2.330293819655522, | |
| "grad_norm": 0.27773727887447136, | |
| "learning_rate": 2.5736064654221808e-05, | |
| "loss": 0.2939, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 2.3343465045592704, | |
| "grad_norm": 0.3034808844409989, | |
| "learning_rate": 2.568166741322048e-05, | |
| "loss": 0.3002, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.3383991894630194, | |
| "grad_norm": 0.3357456105107726, | |
| "learning_rate": 2.56272244128509e-05, | |
| "loss": 0.3285, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 2.342451874366768, | |
| "grad_norm": 0.2632697150242374, | |
| "learning_rate": 2.55727360915895e-05, | |
| "loss": 0.3162, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 2.3465045592705165, | |
| "grad_norm": 0.2854199196850641, | |
| "learning_rate": 2.5518202888277734e-05, | |
| "loss": 0.3176, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 2.3505572441742655, | |
| "grad_norm": 0.29801995922842045, | |
| "learning_rate": 2.5463625242118523e-05, | |
| "loss": 0.3301, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 2.354609929078014, | |
| "grad_norm": 0.28931564276220867, | |
| "learning_rate": 2.5409003592672723e-05, | |
| "loss": 0.3072, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.358662613981763, | |
| "grad_norm": 0.2767863009772188, | |
| "learning_rate": 2.535433837985559e-05, | |
| "loss": 0.3116, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 2.3627152988855116, | |
| "grad_norm": 0.27079692516829984, | |
| "learning_rate": 2.529963004393324e-05, | |
| "loss": 0.307, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 2.3667679837892606, | |
| "grad_norm": 0.2787250215852824, | |
| "learning_rate": 2.524487902551908e-05, | |
| "loss": 0.3279, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 2.370820668693009, | |
| "grad_norm": 0.25912504701376227, | |
| "learning_rate": 2.519008576557029e-05, | |
| "loss": 0.3135, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 2.3748733535967577, | |
| "grad_norm": 0.2666959562025351, | |
| "learning_rate": 2.5135250705384254e-05, | |
| "loss": 0.3221, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 2.3789260385005067, | |
| "grad_norm": 0.28451824746619864, | |
| "learning_rate": 2.5080374286595007e-05, | |
| "loss": 0.326, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 2.382978723404255, | |
| "grad_norm": 0.25767832601899116, | |
| "learning_rate": 2.5025456951169677e-05, | |
| "loss": 0.3264, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 2.387031408308004, | |
| "grad_norm": 0.2829980671554614, | |
| "learning_rate": 2.4970499141404942e-05, | |
| "loss": 0.3277, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 2.3910840932117527, | |
| "grad_norm": 0.2590809631596302, | |
| "learning_rate": 2.491550129992345e-05, | |
| "loss": 0.315, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 2.3951367781155017, | |
| "grad_norm": 0.2628839745929694, | |
| "learning_rate": 2.486046386967024e-05, | |
| "loss": 0.3232, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.3991894630192503, | |
| "grad_norm": 0.28192874923872796, | |
| "learning_rate": 2.4805387293909214e-05, | |
| "loss": 0.319, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 2.403242147922999, | |
| "grad_norm": 0.261394981008923, | |
| "learning_rate": 2.4750272016219552e-05, | |
| "loss": 0.3124, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 2.407294832826748, | |
| "grad_norm": 0.3083102197208763, | |
| "learning_rate": 2.4695118480492114e-05, | |
| "loss": 0.3158, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 2.4113475177304964, | |
| "grad_norm": 0.28406775046012733, | |
| "learning_rate": 2.4639927130925898e-05, | |
| "loss": 0.3219, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 2.4154002026342454, | |
| "grad_norm": 0.32099283690677405, | |
| "learning_rate": 2.458469841202444e-05, | |
| "loss": 0.3277, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 2.419452887537994, | |
| "grad_norm": 0.2776886629242746, | |
| "learning_rate": 2.452943276859226e-05, | |
| "loss": 0.305, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 2.4235055724417425, | |
| "grad_norm": 0.3129534313486284, | |
| "learning_rate": 2.447413064573125e-05, | |
| "loss": 0.3192, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 2.4275582573454915, | |
| "grad_norm": 0.2949852867566288, | |
| "learning_rate": 2.4418792488837095e-05, | |
| "loss": 0.3209, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 2.43161094224924, | |
| "grad_norm": 0.34510973265477923, | |
| "learning_rate": 2.4363418743595713e-05, | |
| "loss": 0.3233, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 2.435663627152989, | |
| "grad_norm": 0.31227471394507644, | |
| "learning_rate": 2.430800985597963e-05, | |
| "loss": 0.3071, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.4397163120567376, | |
| "grad_norm": 0.3553915492372065, | |
| "learning_rate": 2.4252566272244415e-05, | |
| "loss": 0.3182, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 2.443768996960486, | |
| "grad_norm": 0.3320026326332205, | |
| "learning_rate": 2.4197088438925063e-05, | |
| "loss": 0.3243, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 2.447821681864235, | |
| "grad_norm": 0.3117788823937971, | |
| "learning_rate": 2.4141576802832417e-05, | |
| "loss": 0.3035, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 2.4518743667679836, | |
| "grad_norm": 0.3589461356990601, | |
| "learning_rate": 2.408603181104957e-05, | |
| "loss": 0.3158, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 2.4559270516717326, | |
| "grad_norm": 0.31381610860910764, | |
| "learning_rate": 2.4030453910928245e-05, | |
| "loss": 0.3079, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 2.459979736575481, | |
| "grad_norm": 0.3353292363908316, | |
| "learning_rate": 2.397484355008521e-05, | |
| "loss": 0.321, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 2.46403242147923, | |
| "grad_norm": 0.3437871850212179, | |
| "learning_rate": 2.3919201176398662e-05, | |
| "loss": 0.3251, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 2.4680851063829787, | |
| "grad_norm": 0.26780171788908314, | |
| "learning_rate": 2.3863527238004633e-05, | |
| "loss": 0.3242, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 2.4721377912867273, | |
| "grad_norm": 0.3343891224204219, | |
| "learning_rate": 2.380782218329337e-05, | |
| "loss": 0.3062, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 2.4761904761904763, | |
| "grad_norm": 0.27024725161951085, | |
| "learning_rate": 2.3752086460905725e-05, | |
| "loss": 0.3141, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.480243161094225, | |
| "grad_norm": 0.32513260385638726, | |
| "learning_rate": 2.3696320519729544e-05, | |
| "loss": 0.3385, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 2.484295845997974, | |
| "grad_norm": 0.27703958478651913, | |
| "learning_rate": 2.3640524808896045e-05, | |
| "loss": 0.3178, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 2.4883485309017224, | |
| "grad_norm": 0.298713977133974, | |
| "learning_rate": 2.3584699777776222e-05, | |
| "loss": 0.3154, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 2.4924012158054714, | |
| "grad_norm": 0.2841611536378196, | |
| "learning_rate": 2.3528845875977195e-05, | |
| "loss": 0.3323, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 2.49645390070922, | |
| "grad_norm": 0.2947938640613411, | |
| "learning_rate": 2.3472963553338614e-05, | |
| "loss": 0.3132, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 2.5005065856129685, | |
| "grad_norm": 0.3322267752681225, | |
| "learning_rate": 2.341705325992901e-05, | |
| "loss": 0.3273, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 2.5045592705167175, | |
| "grad_norm": 0.26272800490195186, | |
| "learning_rate": 2.336111544604222e-05, | |
| "loss": 0.3104, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 2.508611955420466, | |
| "grad_norm": 0.3093609312781824, | |
| "learning_rate": 2.33051505621937e-05, | |
| "loss": 0.3191, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 2.512664640324215, | |
| "grad_norm": 0.28279578377710796, | |
| "learning_rate": 2.324915905911693e-05, | |
| "loss": 0.3101, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 2.5167173252279635, | |
| "grad_norm": 0.27404034892239826, | |
| "learning_rate": 2.319314138775977e-05, | |
| "loss": 0.3185, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.520770010131712, | |
| "grad_norm": 0.29931252548062, | |
| "learning_rate": 2.3137097999280856e-05, | |
| "loss": 0.3284, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 2.524822695035461, | |
| "grad_norm": 0.2812384477725624, | |
| "learning_rate": 2.308102934504593e-05, | |
| "loss": 0.3176, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 2.5288753799392096, | |
| "grad_norm": 0.3047001306428619, | |
| "learning_rate": 2.3024935876624222e-05, | |
| "loss": 0.2925, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 2.5329280648429586, | |
| "grad_norm": 0.26688578808637126, | |
| "learning_rate": 2.2968818045784813e-05, | |
| "loss": 0.3101, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 2.536980749746707, | |
| "grad_norm": 0.28737307732344064, | |
| "learning_rate": 2.2912676304493006e-05, | |
| "loss": 0.3188, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.5410334346504557, | |
| "grad_norm": 0.2800636440429099, | |
| "learning_rate": 2.2856511104906668e-05, | |
| "loss": 0.3236, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 2.5450861195542047, | |
| "grad_norm": 0.287978351641232, | |
| "learning_rate": 2.2800322899372586e-05, | |
| "loss": 0.335, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 2.5491388044579533, | |
| "grad_norm": 0.3007024362208559, | |
| "learning_rate": 2.2744112140422844e-05, | |
| "loss": 0.3195, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 2.5531914893617023, | |
| "grad_norm": 0.3104388459994633, | |
| "learning_rate": 2.2687879280771177e-05, | |
| "loss": 0.3228, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 2.557244174265451, | |
| "grad_norm": 0.2800771112259633, | |
| "learning_rate": 2.26316247733093e-05, | |
| "loss": 0.3166, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.5612968591691994, | |
| "grad_norm": 0.34826779351463044, | |
| "learning_rate": 2.257534907110328e-05, | |
| "loss": 0.3208, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 2.5653495440729484, | |
| "grad_norm": 0.27544866141515995, | |
| "learning_rate": 2.2519052627389882e-05, | |
| "loss": 0.3213, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 2.569402228976697, | |
| "grad_norm": 0.29603686276623103, | |
| "learning_rate": 2.246273589557294e-05, | |
| "loss": 0.3233, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 2.573454913880446, | |
| "grad_norm": 0.2687241643868818, | |
| "learning_rate": 2.240639932921966e-05, | |
| "loss": 0.3179, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 2.5775075987841944, | |
| "grad_norm": 0.27729111343189555, | |
| "learning_rate": 2.2350043382056995e-05, | |
| "loss": 0.3306, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 2.581560283687943, | |
| "grad_norm": 0.2509570060564368, | |
| "learning_rate": 2.2293668507968015e-05, | |
| "loss": 0.3099, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 2.585612968591692, | |
| "grad_norm": 0.2580980751634586, | |
| "learning_rate": 2.2237275160988186e-05, | |
| "loss": 0.3208, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 2.589665653495441, | |
| "grad_norm": 0.2759536700166203, | |
| "learning_rate": 2.2180863795301787e-05, | |
| "loss": 0.3209, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 2.5937183383991895, | |
| "grad_norm": 0.26636726537280475, | |
| "learning_rate": 2.212443486523819e-05, | |
| "loss": 0.3248, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 2.597771023302938, | |
| "grad_norm": 0.2456330143492669, | |
| "learning_rate": 2.2067988825268243e-05, | |
| "loss": 0.3258, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.601823708206687, | |
| "grad_norm": 0.2972305140135225, | |
| "learning_rate": 2.2011526130000596e-05, | |
| "loss": 0.3181, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 2.6058763931104356, | |
| "grad_norm": 0.25740775006471855, | |
| "learning_rate": 2.1955047234178038e-05, | |
| "loss": 0.3129, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 2.6099290780141846, | |
| "grad_norm": 0.2886020431803275, | |
| "learning_rate": 2.1898552592673825e-05, | |
| "loss": 0.3139, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 2.613981762917933, | |
| "grad_norm": 0.27946994688673366, | |
| "learning_rate": 2.184204266048803e-05, | |
| "loss": 0.3214, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 2.6180344478216817, | |
| "grad_norm": 0.25524211405412833, | |
| "learning_rate": 2.1785517892743887e-05, | |
| "loss": 0.314, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 2.6220871327254307, | |
| "grad_norm": 0.29208836279816774, | |
| "learning_rate": 2.17289787446841e-05, | |
| "loss": 0.3182, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 2.6261398176291793, | |
| "grad_norm": 0.2832074634072403, | |
| "learning_rate": 2.1672425671667198e-05, | |
| "loss": 0.3054, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 2.6301925025329282, | |
| "grad_norm": 0.2740988499711363, | |
| "learning_rate": 2.161585912916385e-05, | |
| "loss": 0.3244, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 2.634245187436677, | |
| "grad_norm": 0.2529095029323145, | |
| "learning_rate": 2.1559279572753214e-05, | |
| "loss": 0.308, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 2.6382978723404253, | |
| "grad_norm": 0.27071069401721337, | |
| "learning_rate": 2.1502687458119268e-05, | |
| "loss": 0.3176, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.6423505572441743, | |
| "grad_norm": 0.2636307989072477, | |
| "learning_rate": 2.1446083241047116e-05, | |
| "loss": 0.3135, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 2.646403242147923, | |
| "grad_norm": 0.2666619578471012, | |
| "learning_rate": 2.1389467377419333e-05, | |
| "loss": 0.3174, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 2.650455927051672, | |
| "grad_norm": 0.27170533188953266, | |
| "learning_rate": 2.133284032321232e-05, | |
| "loss": 0.3085, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 2.6545086119554204, | |
| "grad_norm": 0.30013623216585855, | |
| "learning_rate": 2.1276202534492566e-05, | |
| "loss": 0.3163, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 2.658561296859169, | |
| "grad_norm": 0.2538298822766506, | |
| "learning_rate": 2.121955446741306e-05, | |
| "loss": 0.3255, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.662613981762918, | |
| "grad_norm": 0.2973936781372452, | |
| "learning_rate": 2.1162896578209517e-05, | |
| "loss": 0.3227, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.265097511951897, | |
| "learning_rate": 2.1106229323196813e-05, | |
| "loss": 0.336, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 2.6707193515704155, | |
| "grad_norm": 0.2808003271369898, | |
| "learning_rate": 2.1049553158765214e-05, | |
| "loss": 0.3254, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 2.674772036474164, | |
| "grad_norm": 0.26449055371281965, | |
| "learning_rate": 2.0992868541376764e-05, | |
| "loss": 0.3056, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 2.6788247213779126, | |
| "grad_norm": 0.25367402163967, | |
| "learning_rate": 2.093617592756158e-05, | |
| "loss": 0.3053, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.6828774062816616, | |
| "grad_norm": 0.25444708412736683, | |
| "learning_rate": 2.0879475773914167e-05, | |
| "loss": 0.3074, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 2.6869300911854106, | |
| "grad_norm": 0.2866143943361879, | |
| "learning_rate": 2.082276853708978e-05, | |
| "loss": 0.3084, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 2.690982776089159, | |
| "grad_norm": 0.29316162949818453, | |
| "learning_rate": 2.076605467380071e-05, | |
| "loss": 0.3232, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 2.6950354609929077, | |
| "grad_norm": 0.24817988872190017, | |
| "learning_rate": 2.0709334640812613e-05, | |
| "loss": 0.3176, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 2.6990881458966567, | |
| "grad_norm": 0.28921118513615846, | |
| "learning_rate": 2.0652608894940824e-05, | |
| "loss": 0.2986, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.7031408308004052, | |
| "grad_norm": 0.25328033101339936, | |
| "learning_rate": 2.0595877893046722e-05, | |
| "loss": 0.3174, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 2.7071935157041542, | |
| "grad_norm": 0.30356861795192264, | |
| "learning_rate": 2.0539142092033985e-05, | |
| "loss": 0.3109, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 2.711246200607903, | |
| "grad_norm": 0.2395800783340318, | |
| "learning_rate": 2.048240194884496e-05, | |
| "loss": 0.321, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 2.7152988855116513, | |
| "grad_norm": 0.33503475527370485, | |
| "learning_rate": 2.042565792045695e-05, | |
| "loss": 0.3125, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 2.7193515704154003, | |
| "grad_norm": 0.296053092664007, | |
| "learning_rate": 2.036891046387857e-05, | |
| "loss": 0.3027, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.723404255319149, | |
| "grad_norm": 0.29844694351453116, | |
| "learning_rate": 2.0312160036146036e-05, | |
| "loss": 0.3382, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 2.727456940222898, | |
| "grad_norm": 0.30450560309680474, | |
| "learning_rate": 2.025540709431948e-05, | |
| "loss": 0.3013, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 2.7315096251266464, | |
| "grad_norm": 0.2953276003381534, | |
| "learning_rate": 2.0198652095479298e-05, | |
| "loss": 0.3269, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 2.735562310030395, | |
| "grad_norm": 0.2973472838402754, | |
| "learning_rate": 2.014189549672245e-05, | |
| "loss": 0.3046, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 2.739614994934144, | |
| "grad_norm": 0.31012093815476344, | |
| "learning_rate": 2.0085137755158776e-05, | |
| "loss": 0.319, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.7436676798378925, | |
| "grad_norm": 0.29851982956462453, | |
| "learning_rate": 2.0028379327907327e-05, | |
| "loss": 0.303, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 2.7477203647416415, | |
| "grad_norm": 0.28670577867086294, | |
| "learning_rate": 1.9971620672092676e-05, | |
| "loss": 0.3261, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 2.75177304964539, | |
| "grad_norm": 0.2991719766022081, | |
| "learning_rate": 1.991486224484123e-05, | |
| "loss": 0.317, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 2.7558257345491386, | |
| "grad_norm": 0.29947408776531265, | |
| "learning_rate": 1.985810450327756e-05, | |
| "loss": 0.3245, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 2.7598784194528876, | |
| "grad_norm": 0.2876965086331827, | |
| "learning_rate": 1.9801347904520706e-05, | |
| "loss": 0.32, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.763931104356636, | |
| "grad_norm": 0.28978916438673635, | |
| "learning_rate": 1.974459290568053e-05, | |
| "loss": 0.323, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 2.767983789260385, | |
| "grad_norm": 0.2509183506989627, | |
| "learning_rate": 1.968783996385397e-05, | |
| "loss": 0.3151, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 2.7720364741641337, | |
| "grad_norm": 0.37367080854557766, | |
| "learning_rate": 1.963108953612143e-05, | |
| "loss": 0.3287, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 2.7760891590678822, | |
| "grad_norm": 0.26950827156371626, | |
| "learning_rate": 1.9574342079543056e-05, | |
| "loss": 0.3106, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 2.780141843971631, | |
| "grad_norm": 0.321288993607497, | |
| "learning_rate": 1.9517598051155046e-05, | |
| "loss": 0.3217, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.78419452887538, | |
| "grad_norm": 0.2727608421839769, | |
| "learning_rate": 1.9460857907966025e-05, | |
| "loss": 0.3107, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 2.7882472137791288, | |
| "grad_norm": 0.302666933721235, | |
| "learning_rate": 1.9404122106953285e-05, | |
| "loss": 0.3279, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 2.7922998986828773, | |
| "grad_norm": 0.2731623762343417, | |
| "learning_rate": 1.9347391105059176e-05, | |
| "loss": 0.3185, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 2.7963525835866263, | |
| "grad_norm": 0.3077968572290443, | |
| "learning_rate": 1.92906653591874e-05, | |
| "loss": 0.3391, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 2.800405268490375, | |
| "grad_norm": 0.291307848498414, | |
| "learning_rate": 1.9233945326199295e-05, | |
| "loss": 0.3023, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.804457953394124, | |
| "grad_norm": 0.2743453874139889, | |
| "learning_rate": 1.917723146291022e-05, | |
| "loss": 0.3187, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 2.8085106382978724, | |
| "grad_norm": 0.31346771709935595, | |
| "learning_rate": 1.912052422608584e-05, | |
| "loss": 0.3203, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 2.812563323201621, | |
| "grad_norm": 0.3131699150631023, | |
| "learning_rate": 1.9063824072438428e-05, | |
| "loss": 0.3111, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 2.81661600810537, | |
| "grad_norm": 0.2548998098682518, | |
| "learning_rate": 1.9007131458623246e-05, | |
| "loss": 0.3183, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 2.8206686930091185, | |
| "grad_norm": 0.30085980714007154, | |
| "learning_rate": 1.895044684123479e-05, | |
| "loss": 0.3177, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.8247213779128675, | |
| "grad_norm": 0.25714901297540343, | |
| "learning_rate": 1.8893770676803194e-05, | |
| "loss": 0.3199, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 2.828774062816616, | |
| "grad_norm": 0.28316409587817454, | |
| "learning_rate": 1.8837103421790486e-05, | |
| "loss": 0.3189, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 2.8328267477203646, | |
| "grad_norm": 0.27466453955240794, | |
| "learning_rate": 1.8780445532586952e-05, | |
| "loss": 0.3044, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 2.8368794326241136, | |
| "grad_norm": 0.286397240735338, | |
| "learning_rate": 1.872379746550743e-05, | |
| "loss": 0.3285, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 2.840932117527862, | |
| "grad_norm": 0.2533647215651149, | |
| "learning_rate": 1.866715967678769e-05, | |
| "loss": 0.3179, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.844984802431611, | |
| "grad_norm": 0.2855867034499642, | |
| "learning_rate": 1.861053262258067e-05, | |
| "loss": 0.331, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 2.8490374873353597, | |
| "grad_norm": 0.2689490273902718, | |
| "learning_rate": 1.8553916758952897e-05, | |
| "loss": 0.3402, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 2.853090172239108, | |
| "grad_norm": 0.2822758909060839, | |
| "learning_rate": 1.8497312541880735e-05, | |
| "loss": 0.3157, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 2.857142857142857, | |
| "grad_norm": 0.24349053600867737, | |
| "learning_rate": 1.8440720427246786e-05, | |
| "loss": 0.309, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 2.8611955420466058, | |
| "grad_norm": 0.26320927760220814, | |
| "learning_rate": 1.8384140870836157e-05, | |
| "loss": 0.3171, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.8652482269503547, | |
| "grad_norm": 0.2738340557573272, | |
| "learning_rate": 1.8327574328332806e-05, | |
| "loss": 0.3229, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 2.8693009118541033, | |
| "grad_norm": 0.26255338462316236, | |
| "learning_rate": 1.8271021255315906e-05, | |
| "loss": 0.3251, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 2.873353596757852, | |
| "grad_norm": 0.2590681016270215, | |
| "learning_rate": 1.8214482107256117e-05, | |
| "loss": 0.3163, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 2.877406281661601, | |
| "grad_norm": 0.2735922237957401, | |
| "learning_rate": 1.8157957339511968e-05, | |
| "loss": 0.3103, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 2.8814589665653494, | |
| "grad_norm": 0.25319718349095505, | |
| "learning_rate": 1.8101447407326182e-05, | |
| "loss": 0.3171, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.8855116514690984, | |
| "grad_norm": 0.3093483535601061, | |
| "learning_rate": 1.8044952765821966e-05, | |
| "loss": 0.3017, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 2.889564336372847, | |
| "grad_norm": 0.2738161003770121, | |
| "learning_rate": 1.7988473869999407e-05, | |
| "loss": 0.311, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 2.8936170212765955, | |
| "grad_norm": 0.3025849491712849, | |
| "learning_rate": 1.7932011174731764e-05, | |
| "loss": 0.3087, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 2.8976697061803445, | |
| "grad_norm": 0.29056064473552956, | |
| "learning_rate": 1.7875565134761817e-05, | |
| "loss": 0.3228, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 2.9017223910840935, | |
| "grad_norm": 0.2815530218580791, | |
| "learning_rate": 1.7819136204698226e-05, | |
| "loss": 0.3148, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.905775075987842, | |
| "grad_norm": 0.30990733111360386, | |
| "learning_rate": 1.776272483901182e-05, | |
| "loss": 0.3202, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 2.9098277608915906, | |
| "grad_norm": 0.2819892944378403, | |
| "learning_rate": 1.7706331492031995e-05, | |
| "loss": 0.2989, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 2.9138804457953396, | |
| "grad_norm": 0.26074571667478263, | |
| "learning_rate": 1.764995661794301e-05, | |
| "loss": 0.3235, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 2.917933130699088, | |
| "grad_norm": 0.307692556403157, | |
| "learning_rate": 1.759360067078035e-05, | |
| "loss": 0.3087, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 2.921985815602837, | |
| "grad_norm": 0.2721604456679236, | |
| "learning_rate": 1.7537264104427064e-05, | |
| "loss": 0.3192, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.9260385005065856, | |
| "grad_norm": 0.27082038230095057, | |
| "learning_rate": 1.748094737261012e-05, | |
| "loss": 0.3179, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 2.930091185410334, | |
| "grad_norm": 0.2662688480802123, | |
| "learning_rate": 1.7424650928896726e-05, | |
| "loss": 0.3161, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 2.934143870314083, | |
| "grad_norm": 0.28802404646648105, | |
| "learning_rate": 1.7368375226690712e-05, | |
| "loss": 0.3035, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 2.9381965552178317, | |
| "grad_norm": 0.2867628724013083, | |
| "learning_rate": 1.731212071922883e-05, | |
| "loss": 0.3196, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 2.9422492401215807, | |
| "grad_norm": 0.2600867579152798, | |
| "learning_rate": 1.7255887859577156e-05, | |
| "loss": 0.3113, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.9463019250253293, | |
| "grad_norm": 0.2901990226903036, | |
| "learning_rate": 1.7199677100627427e-05, | |
| "loss": 0.3163, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 2.950354609929078, | |
| "grad_norm": 0.26834088343683243, | |
| "learning_rate": 1.7143488895093343e-05, | |
| "loss": 0.3115, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 2.954407294832827, | |
| "grad_norm": 0.2672873304016795, | |
| "learning_rate": 1.7087323695506994e-05, | |
| "loss": 0.3139, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 2.9584599797365754, | |
| "grad_norm": 0.2661938959763039, | |
| "learning_rate": 1.7031181954215194e-05, | |
| "loss": 0.3103, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 2.9625126646403244, | |
| "grad_norm": 0.2652178948224258, | |
| "learning_rate": 1.6975064123375788e-05, | |
| "loss": 0.3249, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.966565349544073, | |
| "grad_norm": 0.2541143251907282, | |
| "learning_rate": 1.6918970654954084e-05, | |
| "loss": 0.3224, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 2.9706180344478215, | |
| "grad_norm": 0.2771671168654341, | |
| "learning_rate": 1.686290200071915e-05, | |
| "loss": 0.3213, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 2.9746707193515705, | |
| "grad_norm": 0.27561600896334754, | |
| "learning_rate": 1.6806858612240234e-05, | |
| "loss": 0.3196, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 2.978723404255319, | |
| "grad_norm": 0.24076876908410338, | |
| "learning_rate": 1.6750840940883078e-05, | |
| "loss": 0.3141, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 2.982776089159068, | |
| "grad_norm": 0.2597262834883432, | |
| "learning_rate": 1.6694849437806305e-05, | |
| "loss": 0.3108, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.9868287740628165, | |
| "grad_norm": 0.2642615691035671, | |
| "learning_rate": 1.663888455395778e-05, | |
| "loss": 0.3176, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 2.990881458966565, | |
| "grad_norm": 0.2495215562560912, | |
| "learning_rate": 1.6582946740070995e-05, | |
| "loss": 0.3167, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 2.994934143870314, | |
| "grad_norm": 0.25966520599691567, | |
| "learning_rate": 1.6527036446661396e-05, | |
| "loss": 0.3059, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 3.0010131712259374, | |
| "grad_norm": 0.26707035153541575, | |
| "learning_rate": 1.6471154124022818e-05, | |
| "loss": 0.3014, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 3.005065856129686, | |
| "grad_norm": 0.37054529856805907, | |
| "learning_rate": 1.6415300222223788e-05, | |
| "loss": 0.2657, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.0091185410334345, | |
| "grad_norm": 0.3033080941906499, | |
| "learning_rate": 1.6359475191103958e-05, | |
| "loss": 0.2475, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 3.0131712259371835, | |
| "grad_norm": 0.5215864747678008, | |
| "learning_rate": 1.6303679480270466e-05, | |
| "loss": 0.2577, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 3.017223910840932, | |
| "grad_norm": 0.36252736741351743, | |
| "learning_rate": 1.624791353909428e-05, | |
| "loss": 0.2319, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 3.021276595744681, | |
| "grad_norm": 0.40694844811759723, | |
| "learning_rate": 1.619217781670663e-05, | |
| "loss": 0.2589, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 3.0253292806484295, | |
| "grad_norm": 0.35970648541933065, | |
| "learning_rate": 1.6136472761995373e-05, | |
| "loss": 0.2611, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 3.0293819655521785, | |
| "grad_norm": 0.33431133546321407, | |
| "learning_rate": 1.608079882360134e-05, | |
| "loss": 0.2303, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 3.033434650455927, | |
| "grad_norm": 0.3796849192257442, | |
| "learning_rate": 1.60251564499148e-05, | |
| "loss": 0.2465, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 3.0374873353596756, | |
| "grad_norm": 0.4406729282332612, | |
| "learning_rate": 1.596954608907176e-05, | |
| "loss": 0.271, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 3.0415400202634246, | |
| "grad_norm": 0.33700173938923633, | |
| "learning_rate": 1.591396818895043e-05, | |
| "loss": 0.2431, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 3.045592705167173, | |
| "grad_norm": 0.4258988711405886, | |
| "learning_rate": 1.585842319716759e-05, | |
| "loss": 0.251, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.049645390070922, | |
| "grad_norm": 0.3477502246393265, | |
| "learning_rate": 1.5802911561074944e-05, | |
| "loss": 0.2493, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 3.0536980749746707, | |
| "grad_norm": 0.3332850047967658, | |
| "learning_rate": 1.5747433727755595e-05, | |
| "loss": 0.2394, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 3.0577507598784193, | |
| "grad_norm": 0.3499417536664681, | |
| "learning_rate": 1.5691990144020376e-05, | |
| "loss": 0.252, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 3.0618034447821683, | |
| "grad_norm": 0.33951365476429407, | |
| "learning_rate": 1.5636581256404297e-05, | |
| "loss": 0.2517, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 3.065856129685917, | |
| "grad_norm": 0.30816488555095556, | |
| "learning_rate": 1.558120751116291e-05, | |
| "loss": 0.2362, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 3.069908814589666, | |
| "grad_norm": 0.3184952388262607, | |
| "learning_rate": 1.552586935426876e-05, | |
| "loss": 0.2494, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 3.0739614994934144, | |
| "grad_norm": 0.33528765941413813, | |
| "learning_rate": 1.547056723140774e-05, | |
| "loss": 0.2553, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 3.078014184397163, | |
| "grad_norm": 0.3111741568699098, | |
| "learning_rate": 1.5415301587975565e-05, | |
| "loss": 0.2632, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 3.082066869300912, | |
| "grad_norm": 0.30993877066459496, | |
| "learning_rate": 1.536007286907411e-05, | |
| "loss": 0.2448, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 3.0861195542046604, | |
| "grad_norm": 0.2937046215524129, | |
| "learning_rate": 1.5304881519507896e-05, | |
| "loss": 0.2537, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.0901722391084094, | |
| "grad_norm": 0.33709391199463296, | |
| "learning_rate": 1.5249727983780453e-05, | |
| "loss": 0.2589, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 3.094224924012158, | |
| "grad_norm": 0.28493854588437983, | |
| "learning_rate": 1.5194612706090786e-05, | |
| "loss": 0.2544, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 3.098277608915907, | |
| "grad_norm": 0.30744176168766696, | |
| "learning_rate": 1.5139536130329771e-05, | |
| "loss": 0.2393, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 3.1023302938196555, | |
| "grad_norm": 0.31790769684712045, | |
| "learning_rate": 1.508449870007656e-05, | |
| "loss": 0.2578, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 3.106382978723404, | |
| "grad_norm": 0.2938956740562035, | |
| "learning_rate": 1.5029500858595056e-05, | |
| "loss": 0.2622, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 3.110435663627153, | |
| "grad_norm": 0.27099697557920466, | |
| "learning_rate": 1.4974543048830328e-05, | |
| "loss": 0.2435, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 3.1144883485309016, | |
| "grad_norm": 0.27572625733865214, | |
| "learning_rate": 1.4919625713405e-05, | |
| "loss": 0.2452, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 3.1185410334346506, | |
| "grad_norm": 0.26346604981960314, | |
| "learning_rate": 1.4864749294615756e-05, | |
| "loss": 0.2487, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 3.122593718338399, | |
| "grad_norm": 0.30927472070138795, | |
| "learning_rate": 1.4809914234429716e-05, | |
| "loss": 0.2443, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 3.126646403242148, | |
| "grad_norm": 0.2477557856401885, | |
| "learning_rate": 1.4755120974480923e-05, | |
| "loss": 0.257, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 3.1306990881458967, | |
| "grad_norm": 0.2750019957936301, | |
| "learning_rate": 1.4700369956066771e-05, | |
| "loss": 0.2371, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 3.1347517730496453, | |
| "grad_norm": 0.2764247916653975, | |
| "learning_rate": 1.4645661620144413e-05, | |
| "loss": 0.2551, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 3.1388044579533942, | |
| "grad_norm": 0.23935636899141619, | |
| "learning_rate": 1.4590996407327284e-05, | |
| "loss": 0.2541, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 3.142857142857143, | |
| "grad_norm": 0.2573597879742495, | |
| "learning_rate": 1.4536374757881487e-05, | |
| "loss": 0.2598, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 3.146909827760892, | |
| "grad_norm": 0.23608547471848368, | |
| "learning_rate": 1.4481797111722271e-05, | |
| "loss": 0.2373, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.1509625126646403, | |
| "grad_norm": 0.26231738751908235, | |
| "learning_rate": 1.4427263908410507e-05, | |
| "loss": 0.2519, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 3.155015197568389, | |
| "grad_norm": 0.23902041002723293, | |
| "learning_rate": 1.4372775587149108e-05, | |
| "loss": 0.2549, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 3.159067882472138, | |
| "grad_norm": 0.253819928243753, | |
| "learning_rate": 1.4318332586779522e-05, | |
| "loss": 0.2537, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 3.1631205673758864, | |
| "grad_norm": 0.24192290868655536, | |
| "learning_rate": 1.4263935345778202e-05, | |
| "loss": 0.2477, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 3.1671732522796354, | |
| "grad_norm": 0.243486436408247, | |
| "learning_rate": 1.420958430225303e-05, | |
| "loss": 0.257, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 3.171225937183384, | |
| "grad_norm": 0.23094711903045748, | |
| "learning_rate": 1.415527989393985e-05, | |
| "loss": 0.2473, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 3.1752786220871325, | |
| "grad_norm": 0.24919060243998614, | |
| "learning_rate": 1.410102255819891e-05, | |
| "loss": 0.2514, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 3.1793313069908815, | |
| "grad_norm": 0.23706945731180512, | |
| "learning_rate": 1.404681273201131e-05, | |
| "loss": 0.2509, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 3.18338399189463, | |
| "grad_norm": 0.24124745019121094, | |
| "learning_rate": 1.399265085197556e-05, | |
| "loss": 0.2564, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 3.187436676798379, | |
| "grad_norm": 0.249810416084409, | |
| "learning_rate": 1.393853735430398e-05, | |
| "loss": 0.2571, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 3.1914893617021276, | |
| "grad_norm": 0.23614790761729182, | |
| "learning_rate": 1.3884472674819246e-05, | |
| "loss": 0.2586, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 3.1955420466058766, | |
| "grad_norm": 0.24067796141888276, | |
| "learning_rate": 1.3830457248950864e-05, | |
| "loss": 0.2622, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 3.199594731509625, | |
| "grad_norm": 0.23947726986733225, | |
| "learning_rate": 1.377649151173163e-05, | |
| "loss": 0.2519, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 3.2036474164133737, | |
| "grad_norm": 0.2276134135475424, | |
| "learning_rate": 1.3722575897794181e-05, | |
| "loss": 0.2628, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 3.2077001013171227, | |
| "grad_norm": 0.22539888089279891, | |
| "learning_rate": 1.3668710841367472e-05, | |
| "loss": 0.2556, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 3.2117527862208712, | |
| "grad_norm": 0.23303903157621927, | |
| "learning_rate": 1.361489677627324e-05, | |
| "loss": 0.2447, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 3.2158054711246202, | |
| "grad_norm": 0.24562952505124613, | |
| "learning_rate": 1.3561134135922585e-05, | |
| "loss": 0.2688, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 3.219858156028369, | |
| "grad_norm": 0.2792787788668731, | |
| "learning_rate": 1.350742335331241e-05, | |
| "loss": 0.2547, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 3.2239108409321173, | |
| "grad_norm": 0.24454642063390256, | |
| "learning_rate": 1.345376486102198e-05, | |
| "loss": 0.242, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 3.2279635258358663, | |
| "grad_norm": 0.2631757965230207, | |
| "learning_rate": 1.3400159091209414e-05, | |
| "loss": 0.244, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 3.232016210739615, | |
| "grad_norm": 0.25266634725503706, | |
| "learning_rate": 1.3346606475608216e-05, | |
| "loss": 0.2686, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 3.236068895643364, | |
| "grad_norm": 0.25602106907390687, | |
| "learning_rate": 1.3293107445523781e-05, | |
| "loss": 0.2516, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 3.2401215805471124, | |
| "grad_norm": 0.24397857711223694, | |
| "learning_rate": 1.3239662431829949e-05, | |
| "loss": 0.2672, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 3.2441742654508614, | |
| "grad_norm": 0.24622938993201374, | |
| "learning_rate": 1.3186271864965509e-05, | |
| "loss": 0.2505, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 3.24822695035461, | |
| "grad_norm": 0.23578555862890577, | |
| "learning_rate": 1.3132936174930756e-05, | |
| "loss": 0.2537, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.2522796352583585, | |
| "grad_norm": 0.25176576684420776, | |
| "learning_rate": 1.3079655791283995e-05, | |
| "loss": 0.2623, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 3.2563323201621075, | |
| "grad_norm": 0.21892442813728105, | |
| "learning_rate": 1.3026431143138108e-05, | |
| "loss": 0.2356, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 3.260385005065856, | |
| "grad_norm": 0.26074503364066565, | |
| "learning_rate": 1.2973262659157114e-05, | |
| "loss": 0.2528, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 3.264437689969605, | |
| "grad_norm": 0.23360127047883592, | |
| "learning_rate": 1.2920150767552651e-05, | |
| "loss": 0.2444, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 3.2684903748733536, | |
| "grad_norm": 0.2677558876888336, | |
| "learning_rate": 1.2867095896080607e-05, | |
| "loss": 0.2505, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 3.272543059777102, | |
| "grad_norm": 0.2501885566784445, | |
| "learning_rate": 1.2814098472037612e-05, | |
| "loss": 0.2671, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 3.276595744680851, | |
| "grad_norm": 0.22981020970995408, | |
| "learning_rate": 1.276115892225764e-05, | |
| "loss": 0.2626, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 3.2806484295845997, | |
| "grad_norm": 0.2595529833845728, | |
| "learning_rate": 1.2708277673108555e-05, | |
| "loss": 0.252, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 3.2847011144883487, | |
| "grad_norm": 0.2440893128876417, | |
| "learning_rate": 1.2655455150488649e-05, | |
| "loss": 0.2547, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 3.288753799392097, | |
| "grad_norm": 0.23465034795482145, | |
| "learning_rate": 1.2602691779823272e-05, | |
| "loss": 0.2603, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 3.2928064842958458, | |
| "grad_norm": 0.249710039613875, | |
| "learning_rate": 1.2549987986061355e-05, | |
| "loss": 0.2387, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 3.2968591691995948, | |
| "grad_norm": 0.24060753963789408, | |
| "learning_rate": 1.2497344193672005e-05, | |
| "loss": 0.268, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 3.3009118541033433, | |
| "grad_norm": 0.26886310237872957, | |
| "learning_rate": 1.2444760826641092e-05, | |
| "loss": 0.2429, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 3.3049645390070923, | |
| "grad_norm": 0.24455742519923165, | |
| "learning_rate": 1.2392238308467817e-05, | |
| "loss": 0.2375, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 3.309017223910841, | |
| "grad_norm": 0.24496847066529615, | |
| "learning_rate": 1.2339777062161326e-05, | |
| "loss": 0.2561, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 3.31306990881459, | |
| "grad_norm": 0.2637944494398959, | |
| "learning_rate": 1.2287377510237293e-05, | |
| "loss": 0.2594, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 3.3171225937183384, | |
| "grad_norm": 0.23281027021826278, | |
| "learning_rate": 1.2235040074714488e-05, | |
| "loss": 0.2429, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 3.321175278622087, | |
| "grad_norm": 0.27133948985964973, | |
| "learning_rate": 1.2182765177111434e-05, | |
| "loss": 0.238, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 3.325227963525836, | |
| "grad_norm": 0.24332503770075403, | |
| "learning_rate": 1.213055323844297e-05, | |
| "loss": 0.2612, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 3.3292806484295845, | |
| "grad_norm": 0.251008698298789, | |
| "learning_rate": 1.2078404679216864e-05, | |
| "loss": 0.2373, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 0.2503318208256793, | |
| "learning_rate": 1.2026319919430458e-05, | |
| "loss": 0.2556, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 3.337386018237082, | |
| "grad_norm": 0.23712908672866584, | |
| "learning_rate": 1.1974299378567227e-05, | |
| "loss": 0.2481, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 3.341438703140831, | |
| "grad_norm": 0.240890395528491, | |
| "learning_rate": 1.1922343475593462e-05, | |
| "loss": 0.2523, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 3.3454913880445796, | |
| "grad_norm": 0.2432227679477367, | |
| "learning_rate": 1.187045262895488e-05, | |
| "loss": 0.2531, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 3.349544072948328, | |
| "grad_norm": 0.231198351820177, | |
| "learning_rate": 1.1818627256573203e-05, | |
| "loss": 0.2443, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.353596757852077, | |
| "grad_norm": 0.2467027656432149, | |
| "learning_rate": 1.1766867775842864e-05, | |
| "loss": 0.2371, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 3.3576494427558257, | |
| "grad_norm": 0.24923836306897487, | |
| "learning_rate": 1.1715174603627615e-05, | |
| "loss": 0.2519, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 3.3617021276595747, | |
| "grad_norm": 0.2410406862304043, | |
| "learning_rate": 1.1663548156257147e-05, | |
| "loss": 0.2386, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 3.365754812563323, | |
| "grad_norm": 0.23454526958648034, | |
| "learning_rate": 1.161198884952377e-05, | |
| "loss": 0.2471, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 3.3698074974670718, | |
| "grad_norm": 0.24959777767466557, | |
| "learning_rate": 1.1560497098679056e-05, | |
| "loss": 0.2453, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 3.3738601823708207, | |
| "grad_norm": 0.24933352247400797, | |
| "learning_rate": 1.1509073318430479e-05, | |
| "loss": 0.242, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 3.3779128672745693, | |
| "grad_norm": 0.259416390412096, | |
| "learning_rate": 1.1457717922938116e-05, | |
| "loss": 0.2597, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 3.3819655521783183, | |
| "grad_norm": 0.24651285743487417, | |
| "learning_rate": 1.1406431325811233e-05, | |
| "loss": 0.2543, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 3.386018237082067, | |
| "grad_norm": 0.23941582370094036, | |
| "learning_rate": 1.135521394010506e-05, | |
| "loss": 0.2485, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 3.3900709219858154, | |
| "grad_norm": 0.2659250128325504, | |
| "learning_rate": 1.1304066178317367e-05, | |
| "loss": 0.2419, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 3.3941236068895644, | |
| "grad_norm": 0.24136139750263283, | |
| "learning_rate": 1.1252988452385199e-05, | |
| "loss": 0.2514, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 3.398176291793313, | |
| "grad_norm": 0.2398991750219533, | |
| "learning_rate": 1.1201981173681536e-05, | |
| "loss": 0.2434, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 3.402228976697062, | |
| "grad_norm": 0.24418348194553735, | |
| "learning_rate": 1.1151044753011991e-05, | |
| "loss": 0.2567, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 3.4062816616008105, | |
| "grad_norm": 0.24498175604992897, | |
| "learning_rate": 1.1100179600611491e-05, | |
| "loss": 0.2462, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 3.410334346504559, | |
| "grad_norm": 0.23061555419154903, | |
| "learning_rate": 1.1049386126140985e-05, | |
| "loss": 0.2652, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.414387031408308, | |
| "grad_norm": 0.25290747166601246, | |
| "learning_rate": 1.0998664738684128e-05, | |
| "loss": 0.2549, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 3.4184397163120566, | |
| "grad_norm": 0.22462208208609658, | |
| "learning_rate": 1.0948015846744e-05, | |
| "loss": 0.2611, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 3.4224924012158056, | |
| "grad_norm": 0.25210115777818054, | |
| "learning_rate": 1.0897439858239832e-05, | |
| "loss": 0.2526, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 3.426545086119554, | |
| "grad_norm": 0.24844152041847153, | |
| "learning_rate": 1.0846937180503652e-05, | |
| "loss": 0.2433, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 3.430597771023303, | |
| "grad_norm": 0.2631509722650704, | |
| "learning_rate": 1.0796508220277117e-05, | |
| "loss": 0.2416, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 3.4346504559270516, | |
| "grad_norm": 0.24189758676159384, | |
| "learning_rate": 1.0746153383708107e-05, | |
| "loss": 0.2638, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 3.4387031408308006, | |
| "grad_norm": 0.24273136982896407, | |
| "learning_rate": 1.0695873076347579e-05, | |
| "loss": 0.272, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 3.442755825734549, | |
| "grad_norm": 0.22955011590221944, | |
| "learning_rate": 1.0645667703146205e-05, | |
| "loss": 0.2636, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 3.4468085106382977, | |
| "grad_norm": 0.22067648892393424, | |
| "learning_rate": 1.0595537668451161e-05, | |
| "loss": 0.2343, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 3.4508611955420467, | |
| "grad_norm": 0.23496522921232943, | |
| "learning_rate": 1.0545483376002854e-05, | |
| "loss": 0.2526, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.4549138804457953, | |
| "grad_norm": 0.24017651923786498, | |
| "learning_rate": 1.0495505228931676e-05, | |
| "loss": 0.2505, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 3.4589665653495443, | |
| "grad_norm": 0.22826063789629292, | |
| "learning_rate": 1.044560362975474e-05, | |
| "loss": 0.2548, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 3.463019250253293, | |
| "grad_norm": 0.23892781678965233, | |
| "learning_rate": 1.0395778980372695e-05, | |
| "loss": 0.2652, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 3.4670719351570414, | |
| "grad_norm": 0.22333572099300678, | |
| "learning_rate": 1.0346031682066381e-05, | |
| "loss": 0.2375, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 3.4711246200607904, | |
| "grad_norm": 0.23315818809913294, | |
| "learning_rate": 1.0296362135493724e-05, | |
| "loss": 0.2371, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 3.475177304964539, | |
| "grad_norm": 0.2282124427518317, | |
| "learning_rate": 1.0246770740686422e-05, | |
| "loss": 0.2632, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 3.479229989868288, | |
| "grad_norm": 0.22714238442574006, | |
| "learning_rate": 1.0197257897046743e-05, | |
| "loss": 0.2536, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 3.4832826747720365, | |
| "grad_norm": 0.2243033956558672, | |
| "learning_rate": 1.014782400334433e-05, | |
| "loss": 0.2516, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 3.487335359675785, | |
| "grad_norm": 0.22914114803975333, | |
| "learning_rate": 1.009846945771296e-05, | |
| "loss": 0.2391, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 3.491388044579534, | |
| "grad_norm": 0.23949389857089312, | |
| "learning_rate": 1.0049194657647363e-05, | |
| "loss": 0.2529, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 3.4954407294832825, | |
| "grad_norm": 0.2358100803905214, | |
| "learning_rate": 1.0000000000000006e-05, | |
| "loss": 0.2543, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 3.4994934143870315, | |
| "grad_norm": 0.23482621592974048, | |
| "learning_rate": 9.950885880977891e-06, | |
| "loss": 0.2546, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 3.50354609929078, | |
| "grad_norm": 0.26698379807658595, | |
| "learning_rate": 9.901852696139382e-06, | |
| "loss": 0.248, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 3.5075987841945286, | |
| "grad_norm": 0.2398556848462833, | |
| "learning_rate": 9.852900840391027e-06, | |
| "loss": 0.2629, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 3.5116514690982776, | |
| "grad_norm": 0.23526555295396773, | |
| "learning_rate": 9.804030707984313e-06, | |
| "loss": 0.2485, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 3.515704154002026, | |
| "grad_norm": 0.23772688164978428, | |
| "learning_rate": 9.755242692512599e-06, | |
| "loss": 0.2271, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 3.519756838905775, | |
| "grad_norm": 0.23105172251382844, | |
| "learning_rate": 9.70653718690782e-06, | |
| "loss": 0.238, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 3.5238095238095237, | |
| "grad_norm": 0.22793550215673425, | |
| "learning_rate": 9.657914583437454e-06, | |
| "loss": 0.2632, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 3.5278622087132723, | |
| "grad_norm": 0.2404256339474875, | |
| "learning_rate": 9.609375273701246e-06, | |
| "loss": 0.2608, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 3.5319148936170213, | |
| "grad_norm": 0.22587841638196243, | |
| "learning_rate": 9.560919648628133e-06, | |
| "loss": 0.2482, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 3.5359675785207703, | |
| "grad_norm": 0.2250821242468776, | |
| "learning_rate": 9.512548098473047e-06, | |
| "loss": 0.2288, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 3.540020263424519, | |
| "grad_norm": 0.22860862888211847, | |
| "learning_rate": 9.464261012813825e-06, | |
| "loss": 0.2453, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 3.5440729483282674, | |
| "grad_norm": 0.23875316025906732, | |
| "learning_rate": 9.416058780547987e-06, | |
| "loss": 0.2692, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 3.5481256332320164, | |
| "grad_norm": 0.2024060639221084, | |
| "learning_rate": 9.367941789889714e-06, | |
| "loss": 0.2641, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 3.552178318135765, | |
| "grad_norm": 0.22035264195340262, | |
| "learning_rate": 9.319910428366607e-06, | |
| "loss": 0.244, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.556231003039514, | |
| "grad_norm": 0.23243145881964827, | |
| "learning_rate": 9.271965082816667e-06, | |
| "loss": 0.232, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 3.5602836879432624, | |
| "grad_norm": 0.2078298665244665, | |
| "learning_rate": 9.224106139385111e-06, | |
| "loss": 0.2418, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 3.564336372847011, | |
| "grad_norm": 0.22423003589969365, | |
| "learning_rate": 9.176333983521291e-06, | |
| "loss": 0.2477, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 3.56838905775076, | |
| "grad_norm": 0.22258937192272943, | |
| "learning_rate": 9.12864899997558e-06, | |
| "loss": 0.2463, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 3.5724417426545085, | |
| "grad_norm": 0.22006288518914838, | |
| "learning_rate": 9.08105157279628e-06, | |
| "loss": 0.247, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 3.5764944275582575, | |
| "grad_norm": 0.22865628116731315, | |
| "learning_rate": 9.03354208532653e-06, | |
| "loss": 0.2578, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 3.580547112462006, | |
| "grad_norm": 0.23675584051563225, | |
| "learning_rate": 8.986120920201205e-06, | |
| "loss": 0.2374, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 3.5845997973657546, | |
| "grad_norm": 0.23458806647047484, | |
| "learning_rate": 8.938788459343852e-06, | |
| "loss": 0.2574, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 3.5886524822695036, | |
| "grad_norm": 0.23251830444406668, | |
| "learning_rate": 8.8915450839636e-06, | |
| "loss": 0.2423, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 3.592705167173252, | |
| "grad_norm": 0.21683397531275378, | |
| "learning_rate": 8.844391174552116e-06, | |
| "loss": 0.2604, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 3.596757852077001, | |
| "grad_norm": 0.24114125593248117, | |
| "learning_rate": 8.797327110880479e-06, | |
| "loss": 0.242, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 3.6008105369807497, | |
| "grad_norm": 0.24832353903451243, | |
| "learning_rate": 8.750353271996206e-06, | |
| "loss": 0.2648, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 3.6048632218844983, | |
| "grad_norm": 0.23080905554827108, | |
| "learning_rate": 8.703470036220132e-06, | |
| "loss": 0.2499, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 3.6089159067882473, | |
| "grad_norm": 0.2498294327346189, | |
| "learning_rate": 8.656677781143394e-06, | |
| "loss": 0.2478, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 3.612968591691996, | |
| "grad_norm": 0.2384443060752294, | |
| "learning_rate": 8.609976883624377e-06, | |
| "loss": 0.2555, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 3.617021276595745, | |
| "grad_norm": 0.2253518102610019, | |
| "learning_rate": 8.563367719785698e-06, | |
| "loss": 0.2617, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 3.6210739614994933, | |
| "grad_norm": 0.24107070619254534, | |
| "learning_rate": 8.516850665011138e-06, | |
| "loss": 0.2337, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 3.625126646403242, | |
| "grad_norm": 0.26170731398326813, | |
| "learning_rate": 8.47042609394269e-06, | |
| "loss": 0.262, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 3.629179331306991, | |
| "grad_norm": 0.21868534280619686, | |
| "learning_rate": 8.424094380477432e-06, | |
| "loss": 0.2435, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 3.63323201621074, | |
| "grad_norm": 0.22517332349336558, | |
| "learning_rate": 8.37785589776465e-06, | |
| "loss": 0.2556, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 3.6372847011144884, | |
| "grad_norm": 0.22316105968818745, | |
| "learning_rate": 8.331711018202694e-06, | |
| "loss": 0.2649, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 3.641337386018237, | |
| "grad_norm": 0.23233460155002386, | |
| "learning_rate": 8.285660113436104e-06, | |
| "loss": 0.2655, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 3.645390070921986, | |
| "grad_norm": 0.2186489953533202, | |
| "learning_rate": 8.239703554352527e-06, | |
| "loss": 0.2438, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 3.6494427558257345, | |
| "grad_norm": 0.21678438351769364, | |
| "learning_rate": 8.193841711079775e-06, | |
| "loss": 0.2368, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 3.6534954407294835, | |
| "grad_norm": 0.23956065694416673, | |
| "learning_rate": 8.148074952982828e-06, | |
| "loss": 0.2474, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.657548125633232, | |
| "grad_norm": 0.23737857047341232, | |
| "learning_rate": 8.102403648660859e-06, | |
| "loss": 0.27, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 3.6616008105369806, | |
| "grad_norm": 0.2170601526213499, | |
| "learning_rate": 8.056828165944282e-06, | |
| "loss": 0.2475, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 3.6656534954407296, | |
| "grad_norm": 0.2164704769738561, | |
| "learning_rate": 8.011348871891762e-06, | |
| "loss": 0.2623, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 3.669706180344478, | |
| "grad_norm": 0.2522349553988738, | |
| "learning_rate": 7.965966132787287e-06, | |
| "loss": 0.2642, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 3.673758865248227, | |
| "grad_norm": 0.23653314299330108, | |
| "learning_rate": 7.920680314137189e-06, | |
| "loss": 0.2359, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 3.6778115501519757, | |
| "grad_norm": 0.21764116337372533, | |
| "learning_rate": 7.875491780667246e-06, | |
| "loss": 0.2507, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 3.6818642350557242, | |
| "grad_norm": 0.23874535159675875, | |
| "learning_rate": 7.830400896319667e-06, | |
| "loss": 0.2388, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 3.6859169199594732, | |
| "grad_norm": 0.22867076021636784, | |
| "learning_rate": 7.785408024250259e-06, | |
| "loss": 0.2628, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 3.689969604863222, | |
| "grad_norm": 0.22334574831010567, | |
| "learning_rate": 7.74051352682542e-06, | |
| "loss": 0.247, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 3.694022289766971, | |
| "grad_norm": 0.21814963340145405, | |
| "learning_rate": 7.695717765619257e-06, | |
| "loss": 0.2725, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 3.6980749746707193, | |
| "grad_norm": 0.2178246908789961, | |
| "learning_rate": 7.651021101410673e-06, | |
| "loss": 0.2636, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 3.702127659574468, | |
| "grad_norm": 0.23031287800135583, | |
| "learning_rate": 7.606423894180464e-06, | |
| "loss": 0.2654, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 3.706180344478217, | |
| "grad_norm": 0.22962382929698455, | |
| "learning_rate": 7.56192650310839e-06, | |
| "loss": 0.2435, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 3.7102330293819654, | |
| "grad_norm": 0.2098559941350403, | |
| "learning_rate": 7.517529286570349e-06, | |
| "loss": 0.2593, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 3.7142857142857144, | |
| "grad_norm": 0.2337717170277577, | |
| "learning_rate": 7.473232602135387e-06, | |
| "loss": 0.2437, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 3.718338399189463, | |
| "grad_norm": 0.23139609889992876, | |
| "learning_rate": 7.429036806562935e-06, | |
| "loss": 0.2569, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 3.7223910840932115, | |
| "grad_norm": 0.22266747445725357, | |
| "learning_rate": 7.3849422557998455e-06, | |
| "loss": 0.2329, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 3.7264437689969605, | |
| "grad_norm": 0.22731389572554697, | |
| "learning_rate": 7.340949304977567e-06, | |
| "loss": 0.2474, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 3.7304964539007095, | |
| "grad_norm": 0.2196431027407238, | |
| "learning_rate": 7.297058308409282e-06, | |
| "loss": 0.2375, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 3.734549138804458, | |
| "grad_norm": 0.21933121976607284, | |
| "learning_rate": 7.25326961958704e-06, | |
| "loss": 0.2451, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 3.7386018237082066, | |
| "grad_norm": 0.2193137694868286, | |
| "learning_rate": 7.209583591178921e-06, | |
| "loss": 0.2505, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 3.742654508611955, | |
| "grad_norm": 0.22962155854174715, | |
| "learning_rate": 7.1660005750261925e-06, | |
| "loss": 0.2346, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 3.746707193515704, | |
| "grad_norm": 0.25161927456210514, | |
| "learning_rate": 7.1225209221404765e-06, | |
| "loss": 0.2691, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 3.750759878419453, | |
| "grad_norm": 0.22821479265040143, | |
| "learning_rate": 7.079144982700909e-06, | |
| "loss": 0.2514, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 3.7548125633232017, | |
| "grad_norm": 0.2177899837914916, | |
| "learning_rate": 7.0358731060513695e-06, | |
| "loss": 0.2666, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 3.7588652482269502, | |
| "grad_norm": 0.23425882323260724, | |
| "learning_rate": 6.99270564069757e-06, | |
| "loss": 0.2454, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 3.762917933130699, | |
| "grad_norm": 0.22984484205019556, | |
| "learning_rate": 6.949642934304375e-06, | |
| "loss": 0.2486, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 3.7669706180344478, | |
| "grad_norm": 0.2190952174095146, | |
| "learning_rate": 6.906685333692871e-06, | |
| "loss": 0.2329, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 3.7710233029381968, | |
| "grad_norm": 0.21630622540309372, | |
| "learning_rate": 6.86383318483769e-06, | |
| "loss": 0.2424, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 3.7750759878419453, | |
| "grad_norm": 0.22471966441302602, | |
| "learning_rate": 6.821086832864139e-06, | |
| "loss": 0.2554, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 3.779128672745694, | |
| "grad_norm": 0.2282135127911195, | |
| "learning_rate": 6.77844662204546e-06, | |
| "loss": 0.2603, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 3.783181357649443, | |
| "grad_norm": 0.21492286058278282, | |
| "learning_rate": 6.7359128958000455e-06, | |
| "loss": 0.248, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 3.7872340425531914, | |
| "grad_norm": 0.2686621640458993, | |
| "learning_rate": 6.693485996688695e-06, | |
| "loss": 0.249, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 3.7912867274569404, | |
| "grad_norm": 0.218500039706144, | |
| "learning_rate": 6.651166266411801e-06, | |
| "loss": 0.2536, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 3.795339412360689, | |
| "grad_norm": 0.2189935521216615, | |
| "learning_rate": 6.6089540458066725e-06, | |
| "loss": 0.2554, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 3.7993920972644375, | |
| "grad_norm": 0.22603656352927057, | |
| "learning_rate": 6.566849674844711e-06, | |
| "loss": 0.236, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 3.8034447821681865, | |
| "grad_norm": 0.21926310706773428, | |
| "learning_rate": 6.524853492628747e-06, | |
| "loss": 0.2542, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 3.807497467071935, | |
| "grad_norm": 0.21608085738748015, | |
| "learning_rate": 6.4829658373902536e-06, | |
| "loss": 0.2607, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 3.811550151975684, | |
| "grad_norm": 0.21857609397572753, | |
| "learning_rate": 6.441187046486648e-06, | |
| "loss": 0.25, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 3.8156028368794326, | |
| "grad_norm": 0.21107322067368645, | |
| "learning_rate": 6.399517456398567e-06, | |
| "loss": 0.2565, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 3.819655521783181, | |
| "grad_norm": 0.21722856916252445, | |
| "learning_rate": 6.357957402727164e-06, | |
| "loss": 0.2492, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 3.82370820668693, | |
| "grad_norm": 0.21851293421554277, | |
| "learning_rate": 6.316507220191395e-06, | |
| "loss": 0.2476, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 3.8277608915906787, | |
| "grad_norm": 0.2365455132499323, | |
| "learning_rate": 6.275167242625331e-06, | |
| "loss": 0.2339, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 3.8318135764944277, | |
| "grad_norm": 0.22518754657433182, | |
| "learning_rate": 6.233937802975471e-06, | |
| "loss": 0.2592, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 3.835866261398176, | |
| "grad_norm": 0.21953553943601697, | |
| "learning_rate": 6.192819233298046e-06, | |
| "loss": 0.2419, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 3.8399189463019248, | |
| "grad_norm": 0.217562899973498, | |
| "learning_rate": 6.151811864756383e-06, | |
| "loss": 0.2607, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 3.8439716312056738, | |
| "grad_norm": 0.21423451367739377, | |
| "learning_rate": 6.1109160276181655e-06, | |
| "loss": 0.2683, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 3.8480243161094227, | |
| "grad_norm": 0.22924337301258768, | |
| "learning_rate": 6.070132051252868e-06, | |
| "loss": 0.2698, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 3.8520770010131713, | |
| "grad_norm": 0.21323372954653333, | |
| "learning_rate": 6.0294602641290034e-06, | |
| "loss": 0.2492, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 3.85612968591692, | |
| "grad_norm": 0.2220937966117153, | |
| "learning_rate": 5.988900993811575e-06, | |
| "loss": 0.232, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.860182370820669, | |
| "grad_norm": 0.20577746985479678, | |
| "learning_rate": 5.948454566959363e-06, | |
| "loss": 0.2368, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 3.8642350557244174, | |
| "grad_norm": 0.22346418209054447, | |
| "learning_rate": 5.908121309322328e-06, | |
| "loss": 0.2474, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 3.8682877406281664, | |
| "grad_norm": 0.21230776719043867, | |
| "learning_rate": 5.867901545738976e-06, | |
| "loss": 0.2433, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 3.872340425531915, | |
| "grad_norm": 0.22030823735057475, | |
| "learning_rate": 5.827795600133774e-06, | |
| "loss": 0.2566, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 3.8763931104356635, | |
| "grad_norm": 0.22553009086891104, | |
| "learning_rate": 5.787803795514466e-06, | |
| "loss": 0.2521, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 3.8804457953394125, | |
| "grad_norm": 0.21220432267175882, | |
| "learning_rate": 5.747926453969576e-06, | |
| "loss": 0.2443, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 3.884498480243161, | |
| "grad_norm": 0.2189583558635179, | |
| "learning_rate": 5.708163896665708e-06, | |
| "loss": 0.2303, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 3.88855116514691, | |
| "grad_norm": 0.21855865697481855, | |
| "learning_rate": 5.668516443845047e-06, | |
| "loss": 0.2455, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 3.8926038500506586, | |
| "grad_norm": 0.220310114335928, | |
| "learning_rate": 5.6289844148227225e-06, | |
| "loss": 0.26, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 3.896656534954407, | |
| "grad_norm": 0.21060430692750903, | |
| "learning_rate": 5.5895681279842615e-06, | |
| "loss": 0.2605, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 3.900709219858156, | |
| "grad_norm": 0.21354722121523212, | |
| "learning_rate": 5.550267900783019e-06, | |
| "loss": 0.2572, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 3.9047619047619047, | |
| "grad_norm": 0.22417821580806352, | |
| "learning_rate": 5.511084049737623e-06, | |
| "loss": 0.267, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 3.9088145896656536, | |
| "grad_norm": 0.21217884439622287, | |
| "learning_rate": 5.4720168904294215e-06, | |
| "loss": 0.2477, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 3.912867274569402, | |
| "grad_norm": 0.20448732318231558, | |
| "learning_rate": 5.433066737499948e-06, | |
| "loss": 0.2581, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 3.9169199594731507, | |
| "grad_norm": 0.20489544792424377, | |
| "learning_rate": 5.394233904648376e-06, | |
| "loss": 0.2412, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 3.9209726443768997, | |
| "grad_norm": 0.20981051740270204, | |
| "learning_rate": 5.355518704628997e-06, | |
| "loss": 0.2383, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 3.9250253292806483, | |
| "grad_norm": 0.20276616909295075, | |
| "learning_rate": 5.316921449248731e-06, | |
| "loss": 0.2557, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 3.9290780141843973, | |
| "grad_norm": 0.20578972531491474, | |
| "learning_rate": 5.278442449364538e-06, | |
| "loss": 0.2532, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 3.933130699088146, | |
| "grad_norm": 0.22625603786172657, | |
| "learning_rate": 5.240082014881016e-06, | |
| "loss": 0.2233, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 3.9371833839918944, | |
| "grad_norm": 0.20545722996887653, | |
| "learning_rate": 5.201840454747822e-06, | |
| "loss": 0.2615, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 3.9412360688956434, | |
| "grad_norm": 0.207618702360654, | |
| "learning_rate": 5.163718076957223e-06, | |
| "loss": 0.2551, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 3.9452887537993924, | |
| "grad_norm": 0.21464939090890775, | |
| "learning_rate": 5.125715188541609e-06, | |
| "loss": 0.2357, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 3.949341438703141, | |
| "grad_norm": 0.22428172246830821, | |
| "learning_rate": 5.087832095571021e-06, | |
| "loss": 0.2398, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 3.9533941236068895, | |
| "grad_norm": 0.20851723125979466, | |
| "learning_rate": 5.0500691031506766e-06, | |
| "loss": 0.2656, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 3.9574468085106385, | |
| "grad_norm": 0.2124479074830652, | |
| "learning_rate": 5.01242651541854e-06, | |
| "loss": 0.2395, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 3.961499493414387, | |
| "grad_norm": 0.21381408542818298, | |
| "learning_rate": 4.974904635542815e-06, | |
| "loss": 0.2416, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 3.965552178318136, | |
| "grad_norm": 0.21425737026917535, | |
| "learning_rate": 4.937503765719582e-06, | |
| "loss": 0.2541, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 3.9696048632218845, | |
| "grad_norm": 0.21480544496744128, | |
| "learning_rate": 4.900224207170299e-06, | |
| "loss": 0.2507, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 3.973657548125633, | |
| "grad_norm": 0.20591760611694196, | |
| "learning_rate": 4.8630662601394065e-06, | |
| "loss": 0.253, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 3.977710233029382, | |
| "grad_norm": 0.21389821162540285, | |
| "learning_rate": 4.8260302238918995e-06, | |
| "loss": 0.2495, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.9817629179331306, | |
| "grad_norm": 0.21264438578316272, | |
| "learning_rate": 4.789116396710924e-06, | |
| "loss": 0.2377, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 3.9858156028368796, | |
| "grad_norm": 0.21119553197054944, | |
| "learning_rate": 4.752325075895368e-06, | |
| "loss": 0.2426, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 3.989868287740628, | |
| "grad_norm": 0.231235510694373, | |
| "learning_rate": 4.715656557757473e-06, | |
| "loss": 0.2429, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 3.9939209726443767, | |
| "grad_norm": 0.20340249959061651, | |
| "learning_rate": 4.679111137620442e-06, | |
| "loss": 0.2643, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 3.9979736575481257, | |
| "grad_norm": 0.2153206570592999, | |
| "learning_rate": 4.6426891098160585e-06, | |
| "loss": 0.2403, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 4.008105369807497, | |
| "grad_norm": 0.3728250937791261, | |
| "learning_rate": 4.6063907676823474e-06, | |
| "loss": 0.2258, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 4.0121580547112465, | |
| "grad_norm": 0.30904128726278335, | |
| "learning_rate": 4.570216403561141e-06, | |
| "loss": 0.2009, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 4.016210739614995, | |
| "grad_norm": 0.21784797849194495, | |
| "learning_rate": 4.534166308795815e-06, | |
| "loss": 0.2114, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 4.020263424518744, | |
| "grad_norm": 0.3321336042146816, | |
| "learning_rate": 4.498240773728859e-06, | |
| "loss": 0.2116, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 4.024316109422492, | |
| "grad_norm": 0.34297297560903456, | |
| "learning_rate": 4.462440087699609e-06, | |
| "loss": 0.2008, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 4.028368794326241, | |
| "grad_norm": 0.3509474896429846, | |
| "learning_rate": 4.426764539041861e-06, | |
| "loss": 0.2048, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 4.03242147922999, | |
| "grad_norm": 0.25735950667527285, | |
| "learning_rate": 4.391214415081582e-06, | |
| "loss": 0.2136, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 4.036474164133739, | |
| "grad_norm": 0.3053967629921632, | |
| "learning_rate": 4.355790002134579e-06, | |
| "loss": 0.214, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 4.040526849037487, | |
| "grad_norm": 0.3166451500416406, | |
| "learning_rate": 4.320491585504207e-06, | |
| "loss": 0.2135, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 4.044579533941236, | |
| "grad_norm": 0.2812141626834773, | |
| "learning_rate": 4.2853194494790615e-06, | |
| "loss": 0.2059, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 4.048632218844984, | |
| "grad_norm": 0.23200372131336428, | |
| "learning_rate": 4.250273877330691e-06, | |
| "loss": 0.2103, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 4.052684903748734, | |
| "grad_norm": 0.27034476027886584, | |
| "learning_rate": 4.215355151311313e-06, | |
| "loss": 0.2099, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 4.056737588652482, | |
| "grad_norm": 0.29346710888113253, | |
| "learning_rate": 4.180563552651542e-06, | |
| "loss": 0.2074, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 4.060790273556231, | |
| "grad_norm": 0.2663671634063893, | |
| "learning_rate": 4.145899361558147e-06, | |
| "loss": 0.2142, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 4.0648429584599794, | |
| "grad_norm": 0.22971792576929354, | |
| "learning_rate": 4.111362857211738e-06, | |
| "loss": 0.2145, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.068895643363729, | |
| "grad_norm": 0.23528596171134777, | |
| "learning_rate": 4.076954317764592e-06, | |
| "loss": 0.2046, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 4.072948328267477, | |
| "grad_norm": 0.2519359433461938, | |
| "learning_rate": 4.042674020338335e-06, | |
| "loss": 0.2094, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 4.077001013171226, | |
| "grad_norm": 0.2520671086794967, | |
| "learning_rate": 4.0085222410217835e-06, | |
| "loss": 0.2184, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 4.0810536980749745, | |
| "grad_norm": 0.23739763328061916, | |
| "learning_rate": 3.974499254868674e-06, | |
| "loss": 0.2114, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 4.085106382978723, | |
| "grad_norm": 0.22741884501851345, | |
| "learning_rate": 3.940605335895451e-06, | |
| "loss": 0.2245, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 4.0891590678824725, | |
| "grad_norm": 0.22309298901733415, | |
| "learning_rate": 3.90684075707908e-06, | |
| "loss": 0.2036, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 4.093211752786221, | |
| "grad_norm": 0.23103923014305144, | |
| "learning_rate": 3.8732057903548505e-06, | |
| "loss": 0.2072, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 4.09726443768997, | |
| "grad_norm": 0.23193241294615088, | |
| "learning_rate": 3.8397007066141375e-06, | |
| "loss": 0.2034, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 4.101317122593718, | |
| "grad_norm": 0.2169309859665283, | |
| "learning_rate": 3.806325775702304e-06, | |
| "loss": 0.2, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 4.105369807497467, | |
| "grad_norm": 0.21389465949663686, | |
| "learning_rate": 3.773081266416434e-06, | |
| "loss": 0.2206, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 4.109422492401216, | |
| "grad_norm": 0.22404876275378216, | |
| "learning_rate": 3.739967446503245e-06, | |
| "loss": 0.2064, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 4.113475177304965, | |
| "grad_norm": 0.2023544101253924, | |
| "learning_rate": 3.706984582656894e-06, | |
| "loss": 0.2018, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 4.117527862208713, | |
| "grad_norm": 0.200421004004468, | |
| "learning_rate": 3.6741329405168237e-06, | |
| "loss": 0.2185, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 4.121580547112462, | |
| "grad_norm": 0.22182228307737667, | |
| "learning_rate": 3.641412784665648e-06, | |
| "loss": 0.2317, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 4.12563323201621, | |
| "grad_norm": 0.19818919558319126, | |
| "learning_rate": 3.608824378627005e-06, | |
| "loss": 0.2027, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 4.12968591691996, | |
| "grad_norm": 0.1985490850880915, | |
| "learning_rate": 3.5763679848634337e-06, | |
| "loss": 0.2088, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 4.133738601823708, | |
| "grad_norm": 0.20180083457977893, | |
| "learning_rate": 3.544043864774269e-06, | |
| "loss": 0.1988, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 4.137791286727457, | |
| "grad_norm": 0.19568620255994698, | |
| "learning_rate": 3.5118522786935282e-06, | |
| "loss": 0.204, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 4.141843971631205, | |
| "grad_norm": 0.20630266227360602, | |
| "learning_rate": 3.479793485887819e-06, | |
| "loss": 0.1941, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 4.145896656534954, | |
| "grad_norm": 0.2037080804968905, | |
| "learning_rate": 3.4478677445542653e-06, | |
| "loss": 0.2116, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 4.149949341438703, | |
| "grad_norm": 0.20593868223267545, | |
| "learning_rate": 3.4160753118183767e-06, | |
| "loss": 0.2072, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 4.154002026342452, | |
| "grad_norm": 0.2033217112828251, | |
| "learning_rate": 3.3844164437320527e-06, | |
| "loss": 0.2197, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 4.1580547112462005, | |
| "grad_norm": 0.19970293436571815, | |
| "learning_rate": 3.3528913952714558e-06, | |
| "loss": 0.1896, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 4.162107396149949, | |
| "grad_norm": 0.20327539946472403, | |
| "learning_rate": 3.321500420335e-06, | |
| "loss": 0.2167, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 4.1661600810536985, | |
| "grad_norm": 0.1938679368329323, | |
| "learning_rate": 3.290243771741275e-06, | |
| "loss": 0.2175, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.170212765957447, | |
| "grad_norm": 0.19329969246069342, | |
| "learning_rate": 3.2591217012270325e-06, | |
| "loss": 0.2184, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 4.174265450861196, | |
| "grad_norm": 0.20212435343080457, | |
| "learning_rate": 3.228134459445149e-06, | |
| "loss": 0.2068, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 4.178318135764944, | |
| "grad_norm": 0.20064327350110964, | |
| "learning_rate": 3.1972822959626205e-06, | |
| "loss": 0.1989, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 4.182370820668693, | |
| "grad_norm": 0.19150701135778955, | |
| "learning_rate": 3.166565459258513e-06, | |
| "loss": 0.1999, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 4.186423505572442, | |
| "grad_norm": 0.1928450395681863, | |
| "learning_rate": 3.1359841967220193e-06, | |
| "loss": 0.2058, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 4.190476190476191, | |
| "grad_norm": 0.1973148587715577, | |
| "learning_rate": 3.105538754650419e-06, | |
| "loss": 0.2127, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 4.194528875379939, | |
| "grad_norm": 0.19174368175367595, | |
| "learning_rate": 3.07522937824712e-06, | |
| "loss": 0.2129, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 4.198581560283688, | |
| "grad_norm": 0.20238075852186863, | |
| "learning_rate": 3.0450563116196697e-06, | |
| "loss": 0.2112, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 4.202634245187436, | |
| "grad_norm": 0.19124382566256395, | |
| "learning_rate": 3.0150197977778008e-06, | |
| "loss": 0.2135, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 4.206686930091186, | |
| "grad_norm": 0.1973237328597331, | |
| "learning_rate": 2.985120078631465e-06, | |
| "loss": 0.2027, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 4.210739614994934, | |
| "grad_norm": 0.19117467376375802, | |
| "learning_rate": 2.9553573949888893e-06, | |
| "loss": 0.2029, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 4.214792299898683, | |
| "grad_norm": 0.20305641865816706, | |
| "learning_rate": 2.9257319865546384e-06, | |
| "loss": 0.1963, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 4.218844984802431, | |
| "grad_norm": 0.1986742312934355, | |
| "learning_rate": 2.896244091927678e-06, | |
| "loss": 0.2287, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 4.22289766970618, | |
| "grad_norm": 0.2009870647153934, | |
| "learning_rate": 2.8668939485994584e-06, | |
| "loss": 0.2224, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 4.226950354609929, | |
| "grad_norm": 0.19619428025142824, | |
| "learning_rate": 2.837681792951994e-06, | |
| "loss": 0.1976, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 4.231003039513678, | |
| "grad_norm": 0.18964508033103653, | |
| "learning_rate": 2.808607860255981e-06, | |
| "loss": 0.1911, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 4.2350557244174265, | |
| "grad_norm": 0.190758925886592, | |
| "learning_rate": 2.7796723846688634e-06, | |
| "loss": 0.2156, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 4.239108409321175, | |
| "grad_norm": 0.18600778035965768, | |
| "learning_rate": 2.7508755992329937e-06, | |
| "loss": 0.1996, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 4.243161094224924, | |
| "grad_norm": 0.19505017399681737, | |
| "learning_rate": 2.722217735873718e-06, | |
| "loss": 0.2057, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 4.247213779128673, | |
| "grad_norm": 0.19422053598901418, | |
| "learning_rate": 2.6936990253975315e-06, | |
| "loss": 0.1955, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 4.251266464032422, | |
| "grad_norm": 0.20436434185638153, | |
| "learning_rate": 2.665319697490205e-06, | |
| "loss": 0.1923, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 4.25531914893617, | |
| "grad_norm": 0.20136223376288548, | |
| "learning_rate": 2.637079980714945e-06, | |
| "loss": 0.2191, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 4.259371833839919, | |
| "grad_norm": 0.20681761795948075, | |
| "learning_rate": 2.6089801025105453e-06, | |
| "loss": 0.2103, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 4.263424518743667, | |
| "grad_norm": 0.1901462170960438, | |
| "learning_rate": 2.581020289189571e-06, | |
| "loss": 0.1868, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 4.267477203647417, | |
| "grad_norm": 0.19340391868028298, | |
| "learning_rate": 2.553200765936501e-06, | |
| "loss": 0.2164, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.271529888551165, | |
| "grad_norm": 0.19135391938691965, | |
| "learning_rate": 2.525521756805962e-06, | |
| "loss": 0.2103, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 4.275582573454914, | |
| "grad_norm": 0.198405997695439, | |
| "learning_rate": 2.497983484720885e-06, | |
| "loss": 0.1993, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 4.279635258358662, | |
| "grad_norm": 0.2033980721809447, | |
| "learning_rate": 2.470586171470728e-06, | |
| "loss": 0.2103, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 4.283687943262412, | |
| "grad_norm": 0.19857849664618205, | |
| "learning_rate": 2.4433300377096836e-06, | |
| "loss": 0.2094, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 4.28774062816616, | |
| "grad_norm": 0.19936110827305614, | |
| "learning_rate": 2.4162153029549073e-06, | |
| "loss": 0.2178, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 4.291793313069909, | |
| "grad_norm": 0.19832171206169266, | |
| "learning_rate": 2.3892421855847458e-06, | |
| "loss": 0.216, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 4.295845997973657, | |
| "grad_norm": 0.19884911743067568, | |
| "learning_rate": 2.362410902836978e-06, | |
| "loss": 0.1952, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 4.299898682877406, | |
| "grad_norm": 0.2007977624372082, | |
| "learning_rate": 2.3357216708070653e-06, | |
| "loss": 0.2041, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 4.303951367781155, | |
| "grad_norm": 0.21002593308740947, | |
| "learning_rate": 2.309174704446411e-06, | |
| "loss": 0.2108, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 4.308004052684904, | |
| "grad_norm": 0.20233234883727141, | |
| "learning_rate": 2.2827702175606437e-06, | |
| "loss": 0.2194, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 4.3120567375886525, | |
| "grad_norm": 0.20875491337726676, | |
| "learning_rate": 2.256508422807855e-06, | |
| "loss": 0.1997, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 4.316109422492401, | |
| "grad_norm": 0.19201424314866283, | |
| "learning_rate": 2.230389531696946e-06, | |
| "loss": 0.1937, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 4.32016210739615, | |
| "grad_norm": 0.19463341741017312, | |
| "learning_rate": 2.204413754585857e-06, | |
| "loss": 0.2172, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 4.324214792299899, | |
| "grad_norm": 0.19760874842665693, | |
| "learning_rate": 2.1785813006799406e-06, | |
| "loss": 0.1934, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 4.328267477203648, | |
| "grad_norm": 0.18829165130605652, | |
| "learning_rate": 2.1528923780302224e-06, | |
| "loss": 0.2154, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 4.332320162107396, | |
| "grad_norm": 0.18948279703945883, | |
| "learning_rate": 2.127347193531757e-06, | |
| "loss": 0.1987, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 4.336372847011145, | |
| "grad_norm": 0.1985505163405811, | |
| "learning_rate": 2.101945952921942e-06, | |
| "loss": 0.1943, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 4.340425531914893, | |
| "grad_norm": 0.18911711087478236, | |
| "learning_rate": 2.0766888607788906e-06, | |
| "loss": 0.2102, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 4.344478216818643, | |
| "grad_norm": 0.19828740313968918, | |
| "learning_rate": 2.0515761205197337e-06, | |
| "loss": 0.2218, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 4.348530901722391, | |
| "grad_norm": 0.20695534664422935, | |
| "learning_rate": 2.0266079343990453e-06, | |
| "loss": 0.1962, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 4.35258358662614, | |
| "grad_norm": 0.1977653126198484, | |
| "learning_rate": 2.0017845035071494e-06, | |
| "loss": 0.2141, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 4.356636271529888, | |
| "grad_norm": 0.19417161870374833, | |
| "learning_rate": 1.9771060277685537e-06, | |
| "loss": 0.2111, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 4.360688956433638, | |
| "grad_norm": 0.191014988215718, | |
| "learning_rate": 1.95257270594031e-06, | |
| "loss": 0.2088, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 4.364741641337386, | |
| "grad_norm": 0.1985691409797983, | |
| "learning_rate": 1.9281847356104188e-06, | |
| "loss": 0.2187, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 4.368794326241135, | |
| "grad_norm": 0.199652062834898, | |
| "learning_rate": 1.9039423131962365e-06, | |
| "loss": 0.194, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.372847011144883, | |
| "grad_norm": 0.20177068934576847, | |
| "learning_rate": 1.8798456339429027e-06, | |
| "loss": 0.2086, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 4.376899696048632, | |
| "grad_norm": 0.20000689906150051, | |
| "learning_rate": 1.8558948919217612e-06, | |
| "loss": 0.2058, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 4.380952380952381, | |
| "grad_norm": 0.1883138258055086, | |
| "learning_rate": 1.8320902800287954e-06, | |
| "loss": 0.1964, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 4.38500506585613, | |
| "grad_norm": 0.19110762360476072, | |
| "learning_rate": 1.8084319899830726e-06, | |
| "loss": 0.216, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 4.3890577507598785, | |
| "grad_norm": 0.19005707984964473, | |
| "learning_rate": 1.7849202123252097e-06, | |
| "loss": 0.1985, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 4.393110435663627, | |
| "grad_norm": 0.20287389508641734, | |
| "learning_rate": 1.7615551364158401e-06, | |
| "loss": 0.2136, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 4.397163120567376, | |
| "grad_norm": 0.19037790089695714, | |
| "learning_rate": 1.738336950434061e-06, | |
| "loss": 0.1916, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 4.401215805471125, | |
| "grad_norm": 0.1946732644091577, | |
| "learning_rate": 1.715265841375957e-06, | |
| "loss": 0.2054, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 4.405268490374874, | |
| "grad_norm": 0.19054502031139026, | |
| "learning_rate": 1.6923419950530684e-06, | |
| "loss": 0.2097, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 4.409321175278622, | |
| "grad_norm": 0.19706606866281745, | |
| "learning_rate": 1.6695655960909008e-06, | |
| "loss": 0.1964, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 4.413373860182371, | |
| "grad_norm": 0.1957525413979524, | |
| "learning_rate": 1.646936827927441e-06, | |
| "loss": 0.1963, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 4.417426545086119, | |
| "grad_norm": 0.19561475423764677, | |
| "learning_rate": 1.6244558728116766e-06, | |
| "loss": 0.1978, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 4.421479229989869, | |
| "grad_norm": 0.20182599180151264, | |
| "learning_rate": 1.6021229118021265e-06, | |
| "loss": 0.2011, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 4.425531914893617, | |
| "grad_norm": 0.19722660048385274, | |
| "learning_rate": 1.5799381247653967e-06, | |
| "loss": 0.1875, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 4.429584599797366, | |
| "grad_norm": 0.19694027112744078, | |
| "learning_rate": 1.5579016903747013e-06, | |
| "loss": 0.2061, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 4.433637284701114, | |
| "grad_norm": 0.1918264586351235, | |
| "learning_rate": 1.5360137861084656e-06, | |
| "loss": 0.1923, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 4.437689969604863, | |
| "grad_norm": 0.18967259454215465, | |
| "learning_rate": 1.5142745882488475e-06, | |
| "loss": 0.2097, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 4.441742654508612, | |
| "grad_norm": 0.2020383022827132, | |
| "learning_rate": 1.4926842718803691e-06, | |
| "loss": 0.1736, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 4.445795339412361, | |
| "grad_norm": 0.18814888624594353, | |
| "learning_rate": 1.4712430108884657e-06, | |
| "loss": 0.201, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 4.449848024316109, | |
| "grad_norm": 0.19068207501860646, | |
| "learning_rate": 1.4499509779581078e-06, | |
| "loss": 0.2107, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 4.453900709219858, | |
| "grad_norm": 0.18874718249864889, | |
| "learning_rate": 1.4288083445723988e-06, | |
| "loss": 0.2149, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 4.4579533941236065, | |
| "grad_norm": 0.1957110895163618, | |
| "learning_rate": 1.4078152810112045e-06, | |
| "loss": 0.2003, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 4.462006079027356, | |
| "grad_norm": 0.189796038107309, | |
| "learning_rate": 1.3869719563497697e-06, | |
| "loss": 0.1948, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 4.4660587639311045, | |
| "grad_norm": 0.19176735917656818, | |
| "learning_rate": 1.3662785384573663e-06, | |
| "loss": 0.2175, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 4.470111448834853, | |
| "grad_norm": 0.19715662034346806, | |
| "learning_rate": 1.3457351939959383e-06, | |
| "loss": 0.2135, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.474164133738602, | |
| "grad_norm": 0.19226588896100344, | |
| "learning_rate": 1.3253420884187551e-06, | |
| "loss": 0.2019, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 4.47821681864235, | |
| "grad_norm": 0.18512481255549362, | |
| "learning_rate": 1.3050993859690953e-06, | |
| "loss": 0.2185, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 4.4822695035460995, | |
| "grad_norm": 0.19284862766113575, | |
| "learning_rate": 1.2850072496788869e-06, | |
| "loss": 0.2214, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 4.486322188449848, | |
| "grad_norm": 0.21551522239582485, | |
| "learning_rate": 1.2650658413674434e-06, | |
| "loss": 0.2176, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 4.490374873353597, | |
| "grad_norm": 0.19402188193421202, | |
| "learning_rate": 1.2452753216401226e-06, | |
| "loss": 0.1893, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 4.494427558257345, | |
| "grad_norm": 0.1917087826766247, | |
| "learning_rate": 1.2256358498870503e-06, | |
| "loss": 0.2154, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 4.498480243161095, | |
| "grad_norm": 0.18760416770215219, | |
| "learning_rate": 1.2061475842818337e-06, | |
| "loss": 0.2171, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 4.502532928064843, | |
| "grad_norm": 0.18185090949144245, | |
| "learning_rate": 1.1868106817802816e-06, | |
| "loss": 0.1916, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 4.506585612968592, | |
| "grad_norm": 0.20441368381956507, | |
| "learning_rate": 1.1676252981191482e-06, | |
| "loss": 0.2103, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 4.51063829787234, | |
| "grad_norm": 0.1965939773589765, | |
| "learning_rate": 1.1485915878148823e-06, | |
| "loss": 0.2106, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 4.514690982776089, | |
| "grad_norm": 0.18682411597599913, | |
| "learning_rate": 1.1297097041623584e-06, | |
| "loss": 0.1902, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 4.518743667679838, | |
| "grad_norm": 0.1931919088998367, | |
| "learning_rate": 1.1109797992336847e-06, | |
| "loss": 0.2028, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 4.522796352583587, | |
| "grad_norm": 0.1980367609548855, | |
| "learning_rate": 1.092402023876933e-06, | |
| "loss": 0.1948, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 4.526849037487335, | |
| "grad_norm": 0.19408583336049226, | |
| "learning_rate": 1.0739765277149527e-06, | |
| "loss": 0.2117, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 4.530901722391084, | |
| "grad_norm": 0.1903682459416415, | |
| "learning_rate": 1.0557034591441596e-06, | |
| "loss": 0.2097, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 4.5349544072948325, | |
| "grad_norm": 0.20414370254258263, | |
| "learning_rate": 1.0375829653333324e-06, | |
| "loss": 0.1949, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 4.539007092198582, | |
| "grad_norm": 0.19348321650530392, | |
| "learning_rate": 1.0196151922224385e-06, | |
| "loss": 0.2154, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 4.54305977710233, | |
| "grad_norm": 0.1874101017608361, | |
| "learning_rate": 1.0018002845214526e-06, | |
| "loss": 0.2124, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 4.547112462006079, | |
| "grad_norm": 0.19499493164010348, | |
| "learning_rate": 9.841383857091947e-07, | |
| "loss": 0.1829, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 4.5511651469098275, | |
| "grad_norm": 0.19481865631585998, | |
| "learning_rate": 9.666296380321616e-07, | |
| "loss": 0.2023, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 4.555217831813577, | |
| "grad_norm": 0.18968587097568324, | |
| "learning_rate": 9.492741825034124e-07, | |
| "loss": 0.1953, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 4.5592705167173255, | |
| "grad_norm": 0.19299974167627437, | |
| "learning_rate": 9.320721589013892e-07, | |
| "loss": 0.2191, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 4.563323201621074, | |
| "grad_norm": 0.18961973448883343, | |
| "learning_rate": 9.150237057688339e-07, | |
| "loss": 0.2038, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 4.567375886524823, | |
| "grad_norm": 0.19266302724347453, | |
| "learning_rate": 8.981289604116328e-07, | |
| "loss": 0.2115, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 4.571428571428571, | |
| "grad_norm": 0.19021860534247798, | |
| "learning_rate": 8.813880588977542e-07, | |
| "loss": 0.2356, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.575481256332321, | |
| "grad_norm": 0.19612385497806506, | |
| "learning_rate": 8.648011360561126e-07, | |
| "loss": 0.2076, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 4.579533941236069, | |
| "grad_norm": 0.19293685920679862, | |
| "learning_rate": 8.483683254755037e-07, | |
| "loss": 0.2192, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 4.583586626139818, | |
| "grad_norm": 0.19684515855296317, | |
| "learning_rate": 8.320897595035227e-07, | |
| "loss": 0.2266, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 4.587639311043566, | |
| "grad_norm": 0.19182894109428555, | |
| "learning_rate": 8.159655692455093e-07, | |
| "loss": 0.2005, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 4.591691995947315, | |
| "grad_norm": 0.19830925752775308, | |
| "learning_rate": 7.999958845634648e-07, | |
| "loss": 0.1913, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 4.595744680851064, | |
| "grad_norm": 0.19741074172813292, | |
| "learning_rate": 7.841808340750478e-07, | |
| "loss": 0.217, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 4.599797365754813, | |
| "grad_norm": 0.1929603902382358, | |
| "learning_rate": 7.685205451524869e-07, | |
| "loss": 0.207, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 4.603850050658561, | |
| "grad_norm": 0.18831645088906918, | |
| "learning_rate": 7.530151439216027e-07, | |
| "loss": 0.2132, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 4.60790273556231, | |
| "grad_norm": 0.19143463395416266, | |
| "learning_rate": 7.376647552607675e-07, | |
| "loss": 0.2009, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 4.611955420466058, | |
| "grad_norm": 0.19554234724125033, | |
| "learning_rate": 7.224695027998963e-07, | |
| "loss": 0.2101, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 4.616008105369808, | |
| "grad_norm": 0.19971134156689366, | |
| "learning_rate": 7.07429508919466e-07, | |
| "loss": 0.2134, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 4.620060790273556, | |
| "grad_norm": 0.19846238270089578, | |
| "learning_rate": 6.925448947495206e-07, | |
| "loss": 0.1949, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 4.624113475177305, | |
| "grad_norm": 0.18705080646993685, | |
| "learning_rate": 6.778157801686936e-07, | |
| "loss": 0.2232, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 4.6281661600810535, | |
| "grad_norm": 0.19570571783800375, | |
| "learning_rate": 6.632422838032515e-07, | |
| "loss": 0.2046, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 4.632218844984802, | |
| "grad_norm": 0.19298551318114573, | |
| "learning_rate": 6.488245230261281e-07, | |
| "loss": 0.1947, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 4.6362715298885515, | |
| "grad_norm": 0.1907944686973478, | |
| "learning_rate": 6.345626139559868e-07, | |
| "loss": 0.1986, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 4.6403242147923, | |
| "grad_norm": 0.19483729251781115, | |
| "learning_rate": 6.204566714562866e-07, | |
| "loss": 0.2002, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 4.644376899696049, | |
| "grad_norm": 0.1977996993941166, | |
| "learning_rate": 6.06506809134344e-07, | |
| "loss": 0.2129, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 4.648429584599797, | |
| "grad_norm": 0.19246763770509084, | |
| "learning_rate": 5.927131393404373e-07, | |
| "loss": 0.2194, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 4.652482269503546, | |
| "grad_norm": 0.19446591088691975, | |
| "learning_rate": 5.790757731668817e-07, | |
| "loss": 0.2121, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 4.656534954407295, | |
| "grad_norm": 0.19369794278295946, | |
| "learning_rate": 5.655948204471507e-07, | |
| "loss": 0.2044, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 4.660587639311044, | |
| "grad_norm": 0.18217243850659126, | |
| "learning_rate": 5.522703897549875e-07, | |
| "loss": 0.2284, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 4.664640324214792, | |
| "grad_norm": 0.186049050884629, | |
| "learning_rate": 5.391025884035239e-07, | |
| "loss": 0.204, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 4.668693009118541, | |
| "grad_norm": 0.18647309079630014, | |
| "learning_rate": 5.260915224444207e-07, | |
| "loss": 0.2162, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 4.672745694022289, | |
| "grad_norm": 0.19144320747391075, | |
| "learning_rate": 5.132372966670129e-07, | |
| "loss": 0.1917, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.676798378926039, | |
| "grad_norm": 0.18469325185672433, | |
| "learning_rate": 5.005400145974704e-07, | |
| "loss": 0.2139, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 4.680851063829787, | |
| "grad_norm": 0.18993156970263908, | |
| "learning_rate": 4.879997784979562e-07, | |
| "loss": 0.2072, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 4.684903748733536, | |
| "grad_norm": 0.20247946110170692, | |
| "learning_rate": 4.7561668936580984e-07, | |
| "loss": 0.2138, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 4.688956433637284, | |
| "grad_norm": 0.192968325983954, | |
| "learning_rate": 4.6339084693272306e-07, | |
| "loss": 0.1918, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 4.693009118541033, | |
| "grad_norm": 0.18265404408310934, | |
| "learning_rate": 4.5132234966395847e-07, | |
| "loss": 0.2057, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 4.697061803444782, | |
| "grad_norm": 0.2857042652664039, | |
| "learning_rate": 4.3941129475752795e-07, | |
| "loss": 0.2059, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 4.701114488348531, | |
| "grad_norm": 0.18456616218053676, | |
| "learning_rate": 4.27657778143431e-07, | |
| "loss": 0.207, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 4.7051671732522795, | |
| "grad_norm": 0.18777831995541197, | |
| "learning_rate": 4.1606189448287757e-07, | |
| "loss": 0.1997, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 4.709219858156028, | |
| "grad_norm": 0.19669898133793942, | |
| "learning_rate": 4.046237371675177e-07, | |
| "loss": 0.2117, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 4.713272543059777, | |
| "grad_norm": 0.1950346521590039, | |
| "learning_rate": 3.9334339831869963e-07, | |
| "loss": 0.2161, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 4.717325227963526, | |
| "grad_norm": 0.1881083457858523, | |
| "learning_rate": 3.8222096878671955e-07, | |
| "loss": 0.1945, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 4.721377912867275, | |
| "grad_norm": 0.19118052116727005, | |
| "learning_rate": 3.7125653815009545e-07, | |
| "loss": 0.2088, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 4.725430597771023, | |
| "grad_norm": 0.18778867103665528, | |
| "learning_rate": 3.6045019471484974e-07, | |
| "loss": 0.1939, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 4.729483282674772, | |
| "grad_norm": 0.18489705551243543, | |
| "learning_rate": 3.498020255137813e-07, | |
| "loss": 0.1955, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 4.733535967578521, | |
| "grad_norm": 0.1939601658574213, | |
| "learning_rate": 3.393121163057811e-07, | |
| "loss": 0.204, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 4.73758865248227, | |
| "grad_norm": 0.19446047863448182, | |
| "learning_rate": 3.289805515751399e-07, | |
| "loss": 0.2034, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 4.741641337386018, | |
| "grad_norm": 0.18629664418769673, | |
| "learning_rate": 3.188074145308573e-07, | |
| "loss": 0.2094, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 4.745694022289767, | |
| "grad_norm": 0.20546182458964823, | |
| "learning_rate": 3.087927871059804e-07, | |
| "loss": 0.2091, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 4.749746707193515, | |
| "grad_norm": 0.1923062543043768, | |
| "learning_rate": 2.989367499569418e-07, | |
| "loss": 0.2106, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 4.753799392097265, | |
| "grad_norm": 0.19594033109207476, | |
| "learning_rate": 2.8923938246290917e-07, | |
| "loss": 0.2303, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 4.757852077001013, | |
| "grad_norm": 0.18270183076400698, | |
| "learning_rate": 2.7970076272514804e-07, | |
| "loss": 0.2202, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 4.761904761904762, | |
| "grad_norm": 0.1903407990026114, | |
| "learning_rate": 2.703209675663887e-07, | |
| "loss": 0.207, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 4.76595744680851, | |
| "grad_norm": 0.18590348402507142, | |
| "learning_rate": 2.6110007253021374e-07, | |
| "loss": 0.2156, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 4.77001013171226, | |
| "grad_norm": 0.18704494080998238, | |
| "learning_rate": 2.520381518804471e-07, | |
| "loss": 0.1884, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 4.774062816616008, | |
| "grad_norm": 0.18085971414967364, | |
| "learning_rate": 2.4313527860054585e-07, | |
| "loss": 0.2056, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 4.778115501519757, | |
| "grad_norm": 0.18056537223609462, | |
| "learning_rate": 2.343915243930317e-07, | |
| "loss": 0.2102, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 4.7821681864235055, | |
| "grad_norm": 0.1876618076568033, | |
| "learning_rate": 2.2580695967889367e-07, | |
| "loss": 0.1998, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 4.786220871327254, | |
| "grad_norm": 0.18545830159224283, | |
| "learning_rate": 2.1738165359704189e-07, | |
| "loss": 0.2012, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 4.7902735562310035, | |
| "grad_norm": 0.19322942053004538, | |
| "learning_rate": 2.0911567400373257e-07, | |
| "loss": 0.2089, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 4.794326241134752, | |
| "grad_norm": 0.18727127400337054, | |
| "learning_rate": 2.0100908747202607e-07, | |
| "loss": 0.2086, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 4.798378926038501, | |
| "grad_norm": 0.1831338455005895, | |
| "learning_rate": 1.9306195929125638e-07, | |
| "loss": 0.21, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 4.802431610942249, | |
| "grad_norm": 0.1879561439108234, | |
| "learning_rate": 1.8527435346650247e-07, | |
| "loss": 0.2083, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 4.806484295845998, | |
| "grad_norm": 0.18192533762430668, | |
| "learning_rate": 1.7764633271807108e-07, | |
| "loss": 0.2132, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 4.810536980749747, | |
| "grad_norm": 0.18574057026735344, | |
| "learning_rate": 1.7017795848099262e-07, | |
| "loss": 0.2139, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 4.814589665653496, | |
| "grad_norm": 0.1879428142831634, | |
| "learning_rate": 1.6286929090452596e-07, | |
| "loss": 0.2057, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 4.818642350557244, | |
| "grad_norm": 0.19236838013818375, | |
| "learning_rate": 1.557203888516745e-07, | |
| "loss": 0.2093, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 4.822695035460993, | |
| "grad_norm": 0.1979675598764013, | |
| "learning_rate": 1.487313098987131e-07, | |
| "loss": 0.213, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 4.826747720364741, | |
| "grad_norm": 0.18937495902703516, | |
| "learning_rate": 1.4190211033472402e-07, | |
| "loss": 0.2075, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 4.830800405268491, | |
| "grad_norm": 0.18884392447731224, | |
| "learning_rate": 1.3523284516113955e-07, | |
| "loss": 0.2096, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 4.834853090172239, | |
| "grad_norm": 0.18198325909361177, | |
| "learning_rate": 1.2872356809130682e-07, | |
| "loss": 0.2008, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 4.838905775075988, | |
| "grad_norm": 0.2074270815954563, | |
| "learning_rate": 1.2237433155004807e-07, | |
| "loss": 0.2038, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 4.842958459979736, | |
| "grad_norm": 0.1894010558145964, | |
| "learning_rate": 1.1618518667323886e-07, | |
| "loss": 0.1905, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 4.847011144883485, | |
| "grad_norm": 0.19093533260304005, | |
| "learning_rate": 1.1015618330740385e-07, | |
| "loss": 0.2115, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 4.851063829787234, | |
| "grad_norm": 0.185183929555521, | |
| "learning_rate": 1.042873700093061e-07, | |
| "loss": 0.2177, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 4.855116514690983, | |
| "grad_norm": 0.1886583644093005, | |
| "learning_rate": 9.857879404556291e-08, | |
| "loss": 0.2081, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 4.8591691995947315, | |
| "grad_norm": 0.19051056074522707, | |
| "learning_rate": 9.303050139225722e-08, | |
| "loss": 0.2095, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 4.86322188449848, | |
| "grad_norm": 0.19196747905963105, | |
| "learning_rate": 8.76425367345779e-08, | |
| "loss": 0.1898, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 4.867274569402229, | |
| "grad_norm": 0.19317741643174532, | |
| "learning_rate": 8.241494346644897e-08, | |
| "loss": 0.2048, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 4.871327254305978, | |
| "grad_norm": 0.18633771259917592, | |
| "learning_rate": 7.734776369019204e-08, | |
| "loss": 0.1965, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 4.875379939209727, | |
| "grad_norm": 0.18812123894120664, | |
| "learning_rate": 7.244103821617332e-08, | |
| "loss": 0.2088, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.879432624113475, | |
| "grad_norm": 0.18286652952514226, | |
| "learning_rate": 6.769480656248606e-08, | |
| "loss": 0.2185, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 4.883485309017224, | |
| "grad_norm": 0.1992191198114373, | |
| "learning_rate": 6.310910695462635e-08, | |
| "loss": 0.1892, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 4.887537993920972, | |
| "grad_norm": 0.1889486841805104, | |
| "learning_rate": 5.8683976325191185e-08, | |
| "loss": 0.1948, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 4.891590678824722, | |
| "grad_norm": 0.193303248742088, | |
| "learning_rate": 5.4419450313571984e-08, | |
| "loss": 0.2093, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 4.89564336372847, | |
| "grad_norm": 0.18650648902607253, | |
| "learning_rate": 5.031556326567488e-08, | |
| "loss": 0.1832, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 4.899696048632219, | |
| "grad_norm": 0.19343378517257795, | |
| "learning_rate": 4.637234823364312e-08, | |
| "loss": 0.2109, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 4.903748733535967, | |
| "grad_norm": 0.1786091900452692, | |
| "learning_rate": 4.258983697558838e-08, | |
| "loss": 0.2127, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 4.907801418439716, | |
| "grad_norm": 0.1919981650852317, | |
| "learning_rate": 3.896805995533548e-08, | |
| "loss": 0.2104, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 4.911854103343465, | |
| "grad_norm": 0.19062438685148397, | |
| "learning_rate": 3.550704634218028e-08, | |
| "loss": 0.2015, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 4.915906788247214, | |
| "grad_norm": 0.18292986716529297, | |
| "learning_rate": 3.2206824010647676e-08, | |
| "loss": 0.1841, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 4.919959473150962, | |
| "grad_norm": 0.18834245601602922, | |
| "learning_rate": 2.9067419540278476e-08, | |
| "loss": 0.2012, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 4.924012158054711, | |
| "grad_norm": 0.1937701789924416, | |
| "learning_rate": 2.6088858215400638e-08, | |
| "loss": 0.1913, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 4.92806484295846, | |
| "grad_norm": 0.1829685685382398, | |
| "learning_rate": 2.3271164024940564e-08, | |
| "loss": 0.2014, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 4.932117527862209, | |
| "grad_norm": 0.18894283326369202, | |
| "learning_rate": 2.061435966221881e-08, | |
| "loss": 0.2184, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 4.9361702127659575, | |
| "grad_norm": 0.18859954810829604, | |
| "learning_rate": 1.811846652477245e-08, | |
| "loss": 0.2081, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 4.940222897669706, | |
| "grad_norm": 0.18599732857716977, | |
| "learning_rate": 1.5783504714184106e-08, | |
| "loss": 0.2055, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 4.944275582573455, | |
| "grad_norm": 0.19100573532774665, | |
| "learning_rate": 1.360949303591097e-08, | |
| "loss": 0.2025, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 4.948328267477204, | |
| "grad_norm": 0.18451467281757894, | |
| "learning_rate": 1.1596448999144916e-08, | |
| "loss": 0.21, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 4.9523809523809526, | |
| "grad_norm": 0.18746891190968507, | |
| "learning_rate": 9.744388816668172e-09, | |
| "loss": 0.2141, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 4.956433637284701, | |
| "grad_norm": 0.19289260607143519, | |
| "learning_rate": 8.05332740472009e-09, | |
| "loss": 0.215, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 4.96048632218845, | |
| "grad_norm": 0.1887889849048269, | |
| "learning_rate": 6.523278382872811e-09, | |
| "loss": 0.2148, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 4.964539007092198, | |
| "grad_norm": 0.18658048585228024, | |
| "learning_rate": 5.15425407393133e-09, | |
| "loss": 0.1945, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 4.968591691995948, | |
| "grad_norm": 0.19891216809594414, | |
| "learning_rate": 3.94626550383137e-09, | |
| "loss": 0.2033, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 4.972644376899696, | |
| "grad_norm": 0.18188505672344066, | |
| "learning_rate": 2.899322401546112e-09, | |
| "loss": 0.2147, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 4.976697061803445, | |
| "grad_norm": 0.19213171080056604, | |
| "learning_rate": 2.013433199010706e-09, | |
| "loss": 0.2094, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 4.980749746707193, | |
| "grad_norm": 0.19302533292740526, | |
| "learning_rate": 1.2886050310556563e-09, | |
| "loss": 0.2078, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 4.984802431610943, | |
| "grad_norm": 0.18578150037746954, | |
| "learning_rate": 7.248437353468695e-10, | |
| "loss": 0.2015, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 4.988855116514691, | |
| "grad_norm": 0.19401346923047483, | |
| "learning_rate": 3.221538523412449e-10, | |
| "loss": 0.1885, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 4.99290780141844, | |
| "grad_norm": 0.18576483671011168, | |
| "learning_rate": 8.053862524670663e-11, | |
| "loss": 0.211, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 4.996960486322188, | |
| "grad_norm": 0.18530246336028122, | |
| "learning_rate": 0.0, | |
| "loss": 0.2009, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 4.996960486322188, | |
| "step": 1230, | |
| "total_flos": 4.316481383850574e+18, | |
| "train_loss": 0.04117072325169555, | |
| "train_runtime": 10235.9458, | |
| "train_samples_per_second": 15.425, | |
| "train_steps_per_second": 0.12 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1230, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.316481383850574e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |