| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 0, | |
| "global_step": 246, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0040650406504065045, | |
| "grad_norm": 1.3748667240142822, | |
| "learning_rate": 1e-05, | |
| "loss": 2.5758, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008130081300813009, | |
| "grad_norm": 1.2784767150878906, | |
| "learning_rate": 9.959349593495936e-06, | |
| "loss": 2.5216, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.012195121951219513, | |
| "grad_norm": 1.2706096172332764, | |
| "learning_rate": 9.91869918699187e-06, | |
| "loss": 2.5553, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.016260162601626018, | |
| "grad_norm": 1.2058460712432861, | |
| "learning_rate": 9.878048780487805e-06, | |
| "loss": 2.5158, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02032520325203252, | |
| "grad_norm": 1.1286356449127197, | |
| "learning_rate": 9.837398373983741e-06, | |
| "loss": 2.4486, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.024390243902439025, | |
| "grad_norm": 1.125715732574463, | |
| "learning_rate": 9.796747967479675e-06, | |
| "loss": 2.4741, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.028455284552845527, | |
| "grad_norm": 1.089099407196045, | |
| "learning_rate": 9.756097560975611e-06, | |
| "loss": 2.4565, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.032520325203252036, | |
| "grad_norm": 1.068673014640808, | |
| "learning_rate": 9.715447154471546e-06, | |
| "loss": 2.4487, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.036585365853658534, | |
| "grad_norm": 0.9413436055183411, | |
| "learning_rate": 9.67479674796748e-06, | |
| "loss": 2.3395, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04065040650406504, | |
| "grad_norm": 0.8963584303855896, | |
| "learning_rate": 9.634146341463415e-06, | |
| "loss": 2.3072, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.044715447154471545, | |
| "grad_norm": 0.9504267573356628, | |
| "learning_rate": 9.59349593495935e-06, | |
| "loss": 2.4042, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04878048780487805, | |
| "grad_norm": 0.8742660284042358, | |
| "learning_rate": 9.552845528455286e-06, | |
| "loss": 2.3433, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.052845528455284556, | |
| "grad_norm": 0.8436909914016724, | |
| "learning_rate": 9.51219512195122e-06, | |
| "loss": 2.2921, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.056910569105691054, | |
| "grad_norm": 0.8690633177757263, | |
| "learning_rate": 9.471544715447156e-06, | |
| "loss": 2.3208, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06097560975609756, | |
| "grad_norm": 0.8250911235809326, | |
| "learning_rate": 9.43089430894309e-06, | |
| "loss": 2.2882, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06504065040650407, | |
| "grad_norm": 0.8128834962844849, | |
| "learning_rate": 9.390243902439025e-06, | |
| "loss": 2.2526, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06910569105691057, | |
| "grad_norm": 0.8213804960250854, | |
| "learning_rate": 9.34959349593496e-06, | |
| "loss": 2.2962, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07317073170731707, | |
| "grad_norm": 0.7577650547027588, | |
| "learning_rate": 9.308943089430895e-06, | |
| "loss": 2.2449, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.07723577235772358, | |
| "grad_norm": 0.790264368057251, | |
| "learning_rate": 9.268292682926831e-06, | |
| "loss": 2.2296, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08130081300813008, | |
| "grad_norm": 0.7232661247253418, | |
| "learning_rate": 9.227642276422764e-06, | |
| "loss": 2.1508, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08536585365853659, | |
| "grad_norm": 0.7576490640640259, | |
| "learning_rate": 9.1869918699187e-06, | |
| "loss": 2.2099, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08943089430894309, | |
| "grad_norm": 0.7341296672821045, | |
| "learning_rate": 9.146341463414635e-06, | |
| "loss": 2.1892, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.09349593495934959, | |
| "grad_norm": 0.7156996130943298, | |
| "learning_rate": 9.10569105691057e-06, | |
| "loss": 2.1675, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0975609756097561, | |
| "grad_norm": 0.6440656185150146, | |
| "learning_rate": 9.065040650406505e-06, | |
| "loss": 2.0993, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.1016260162601626, | |
| "grad_norm": 0.6809216141700745, | |
| "learning_rate": 9.02439024390244e-06, | |
| "loss": 2.1675, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10569105691056911, | |
| "grad_norm": 0.6825043559074402, | |
| "learning_rate": 8.983739837398374e-06, | |
| "loss": 2.1462, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.10975609756097561, | |
| "grad_norm": 0.6550068855285645, | |
| "learning_rate": 8.94308943089431e-06, | |
| "loss": 2.1678, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11382113821138211, | |
| "grad_norm": 0.6182063221931458, | |
| "learning_rate": 8.902439024390244e-06, | |
| "loss": 2.136, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.11788617886178862, | |
| "grad_norm": 0.6627913117408752, | |
| "learning_rate": 8.86178861788618e-06, | |
| "loss": 2.0555, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.12195121951219512, | |
| "grad_norm": 0.6257705688476562, | |
| "learning_rate": 8.821138211382113e-06, | |
| "loss": 2.1095, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12601626016260162, | |
| "grad_norm": 0.5838558673858643, | |
| "learning_rate": 8.78048780487805e-06, | |
| "loss": 2.0794, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.13008130081300814, | |
| "grad_norm": 0.6288259029388428, | |
| "learning_rate": 8.739837398373985e-06, | |
| "loss": 2.1067, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.13414634146341464, | |
| "grad_norm": 0.5768400430679321, | |
| "learning_rate": 8.69918699186992e-06, | |
| "loss": 2.0634, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.13821138211382114, | |
| "grad_norm": 0.6065109372138977, | |
| "learning_rate": 8.658536585365854e-06, | |
| "loss": 2.0407, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.14227642276422764, | |
| "grad_norm": 0.588671088218689, | |
| "learning_rate": 8.617886178861789e-06, | |
| "loss": 2.0634, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.14634146341463414, | |
| "grad_norm": 0.5733758807182312, | |
| "learning_rate": 8.577235772357724e-06, | |
| "loss": 2.044, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.15040650406504066, | |
| "grad_norm": 0.5910205841064453, | |
| "learning_rate": 8.536585365853658e-06, | |
| "loss": 2.0375, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.15447154471544716, | |
| "grad_norm": 0.5399029850959778, | |
| "learning_rate": 8.495934959349595e-06, | |
| "loss": 2.0321, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.15853658536585366, | |
| "grad_norm": 0.6022946238517761, | |
| "learning_rate": 8.45528455284553e-06, | |
| "loss": 2.0283, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.16260162601626016, | |
| "grad_norm": 0.5221655368804932, | |
| "learning_rate": 8.414634146341464e-06, | |
| "loss": 1.9423, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.5715436339378357, | |
| "learning_rate": 8.373983739837399e-06, | |
| "loss": 1.896, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.17073170731707318, | |
| "grad_norm": 0.5900686979293823, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 1.9766, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.17479674796747968, | |
| "grad_norm": 0.4903549551963806, | |
| "learning_rate": 8.292682926829268e-06, | |
| "loss": 1.9507, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.17886178861788618, | |
| "grad_norm": 0.511839747428894, | |
| "learning_rate": 8.252032520325203e-06, | |
| "loss": 1.8829, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.18292682926829268, | |
| "grad_norm": 0.6022465229034424, | |
| "learning_rate": 8.21138211382114e-06, | |
| "loss": 2.0382, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.18699186991869918, | |
| "grad_norm": 0.5093381404876709, | |
| "learning_rate": 8.170731707317073e-06, | |
| "loss": 1.9458, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1910569105691057, | |
| "grad_norm": 0.5268444418907166, | |
| "learning_rate": 8.130081300813009e-06, | |
| "loss": 1.9686, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.1951219512195122, | |
| "grad_norm": 0.47973573207855225, | |
| "learning_rate": 8.089430894308944e-06, | |
| "loss": 1.9357, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1991869918699187, | |
| "grad_norm": 0.4572773575782776, | |
| "learning_rate": 8.048780487804879e-06, | |
| "loss": 1.9086, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.2032520325203252, | |
| "grad_norm": 0.4701896011829376, | |
| "learning_rate": 8.008130081300813e-06, | |
| "loss": 1.9454, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2073170731707317, | |
| "grad_norm": 0.49787646532058716, | |
| "learning_rate": 7.967479674796748e-06, | |
| "loss": 1.9274, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.21138211382113822, | |
| "grad_norm": 0.450969398021698, | |
| "learning_rate": 7.926829268292685e-06, | |
| "loss": 1.9194, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.21544715447154472, | |
| "grad_norm": 0.4827839732170105, | |
| "learning_rate": 7.886178861788618e-06, | |
| "loss": 1.8947, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.21951219512195122, | |
| "grad_norm": 0.47519227862358093, | |
| "learning_rate": 7.845528455284554e-06, | |
| "loss": 1.919, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.22357723577235772, | |
| "grad_norm": 0.4576883614063263, | |
| "learning_rate": 7.804878048780489e-06, | |
| "loss": 1.9011, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22764227642276422, | |
| "grad_norm": 0.4596790671348572, | |
| "learning_rate": 7.764227642276424e-06, | |
| "loss": 1.9009, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.23170731707317074, | |
| "grad_norm": 0.46299463510513306, | |
| "learning_rate": 7.723577235772358e-06, | |
| "loss": 1.9214, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.23577235772357724, | |
| "grad_norm": 0.42920413613319397, | |
| "learning_rate": 7.682926829268293e-06, | |
| "loss": 1.8673, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.23983739837398374, | |
| "grad_norm": 0.454936146736145, | |
| "learning_rate": 7.64227642276423e-06, | |
| "loss": 1.8651, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.24390243902439024, | |
| "grad_norm": 0.4699499309062958, | |
| "learning_rate": 7.601626016260163e-06, | |
| "loss": 1.8484, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.24796747967479674, | |
| "grad_norm": 0.4237484335899353, | |
| "learning_rate": 7.560975609756098e-06, | |
| "loss": 1.8947, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.25203252032520324, | |
| "grad_norm": 0.4287896156311035, | |
| "learning_rate": 7.520325203252034e-06, | |
| "loss": 1.8491, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.25609756097560976, | |
| "grad_norm": 0.45891255140304565, | |
| "learning_rate": 7.4796747967479676e-06, | |
| "loss": 1.8436, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.2601626016260163, | |
| "grad_norm": 0.4400513470172882, | |
| "learning_rate": 7.439024390243903e-06, | |
| "loss": 1.8479, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.26422764227642276, | |
| "grad_norm": 0.42221692204475403, | |
| "learning_rate": 7.398373983739838e-06, | |
| "loss": 1.853, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.2682926829268293, | |
| "grad_norm": 0.4163649380207062, | |
| "learning_rate": 7.357723577235773e-06, | |
| "loss": 1.8083, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.27235772357723576, | |
| "grad_norm": 0.4221528172492981, | |
| "learning_rate": 7.317073170731707e-06, | |
| "loss": 1.8727, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.2764227642276423, | |
| "grad_norm": 0.4102783799171448, | |
| "learning_rate": 7.276422764227643e-06, | |
| "loss": 1.8416, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2804878048780488, | |
| "grad_norm": 0.42232567071914673, | |
| "learning_rate": 7.2357723577235786e-06, | |
| "loss": 1.8362, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2845528455284553, | |
| "grad_norm": 0.4951733946800232, | |
| "learning_rate": 7.1951219512195125e-06, | |
| "loss": 1.8302, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2886178861788618, | |
| "grad_norm": 0.44171613454818726, | |
| "learning_rate": 7.154471544715448e-06, | |
| "loss": 1.8404, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.2926829268292683, | |
| "grad_norm": 0.4147375524044037, | |
| "learning_rate": 7.113821138211383e-06, | |
| "loss": 1.7993, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2967479674796748, | |
| "grad_norm": 0.41523224115371704, | |
| "learning_rate": 7.0731707317073175e-06, | |
| "loss": 1.7972, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.3008130081300813, | |
| "grad_norm": 0.4788653552532196, | |
| "learning_rate": 7.032520325203252e-06, | |
| "loss": 1.7999, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.3048780487804878, | |
| "grad_norm": 0.41081106662750244, | |
| "learning_rate": 6.991869918699188e-06, | |
| "loss": 1.8774, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3089430894308943, | |
| "grad_norm": 0.45427748560905457, | |
| "learning_rate": 6.951219512195122e-06, | |
| "loss": 1.7976, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3130081300813008, | |
| "grad_norm": 0.4211747944355011, | |
| "learning_rate": 6.910569105691057e-06, | |
| "loss": 1.8265, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3170731707317073, | |
| "grad_norm": 0.482440710067749, | |
| "learning_rate": 6.869918699186993e-06, | |
| "loss": 1.8224, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.32113821138211385, | |
| "grad_norm": 0.4471375346183777, | |
| "learning_rate": 6.829268292682928e-06, | |
| "loss": 1.8346, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3252032520325203, | |
| "grad_norm": 0.43011748790740967, | |
| "learning_rate": 6.788617886178862e-06, | |
| "loss": 1.8136, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.32926829268292684, | |
| "grad_norm": 0.41438254714012146, | |
| "learning_rate": 6.747967479674797e-06, | |
| "loss": 1.7561, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.43242278695106506, | |
| "learning_rate": 6.707317073170733e-06, | |
| "loss": 1.8202, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.33739837398373984, | |
| "grad_norm": 0.4559822380542755, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.7962, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.34146341463414637, | |
| "grad_norm": 0.41565415263175964, | |
| "learning_rate": 6.626016260162602e-06, | |
| "loss": 1.7748, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.34552845528455284, | |
| "grad_norm": 0.3826720416545868, | |
| "learning_rate": 6.585365853658538e-06, | |
| "loss": 1.7771, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.34959349593495936, | |
| "grad_norm": 0.4084075093269348, | |
| "learning_rate": 6.544715447154472e-06, | |
| "loss": 1.7623, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.35365853658536583, | |
| "grad_norm": 0.4102393686771393, | |
| "learning_rate": 6.504065040650407e-06, | |
| "loss": 1.777, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.35772357723577236, | |
| "grad_norm": 0.47018152475357056, | |
| "learning_rate": 6.463414634146342e-06, | |
| "loss": 1.7526, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3617886178861789, | |
| "grad_norm": 0.4740968942642212, | |
| "learning_rate": 6.422764227642278e-06, | |
| "loss": 1.8063, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.36585365853658536, | |
| "grad_norm": 0.40948763489723206, | |
| "learning_rate": 6.3821138211382115e-06, | |
| "loss": 1.7836, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3699186991869919, | |
| "grad_norm": 0.4181380271911621, | |
| "learning_rate": 6.341463414634147e-06, | |
| "loss": 1.8029, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.37398373983739835, | |
| "grad_norm": 0.478249728679657, | |
| "learning_rate": 6.300813008130082e-06, | |
| "loss": 1.7783, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.3780487804878049, | |
| "grad_norm": 0.3988848328590393, | |
| "learning_rate": 6.260162601626017e-06, | |
| "loss": 1.7572, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3821138211382114, | |
| "grad_norm": 0.4052281379699707, | |
| "learning_rate": 6.219512195121951e-06, | |
| "loss": 1.8084, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.3861788617886179, | |
| "grad_norm": 0.5780310034751892, | |
| "learning_rate": 6.178861788617887e-06, | |
| "loss": 1.7885, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.3902439024390244, | |
| "grad_norm": 0.3908810019493103, | |
| "learning_rate": 6.138211382113821e-06, | |
| "loss": 1.7261, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.3943089430894309, | |
| "grad_norm": 0.44484543800354004, | |
| "learning_rate": 6.0975609756097564e-06, | |
| "loss": 1.7299, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.3983739837398374, | |
| "grad_norm": 0.48491889238357544, | |
| "learning_rate": 6.056910569105692e-06, | |
| "loss": 1.7355, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4024390243902439, | |
| "grad_norm": 0.4945378005504608, | |
| "learning_rate": 6.016260162601627e-06, | |
| "loss": 1.7955, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4065040650406504, | |
| "grad_norm": 0.44714319705963135, | |
| "learning_rate": 5.9756097560975615e-06, | |
| "loss": 1.7603, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4105691056910569, | |
| "grad_norm": 0.4738294184207916, | |
| "learning_rate": 5.934959349593496e-06, | |
| "loss": 1.7065, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4146341463414634, | |
| "grad_norm": 0.429187536239624, | |
| "learning_rate": 5.894308943089432e-06, | |
| "loss": 1.7831, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.4186991869918699, | |
| "grad_norm": 0.47387829422950745, | |
| "learning_rate": 5.853658536585366e-06, | |
| "loss": 1.7591, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.42276422764227645, | |
| "grad_norm": 0.4267633557319641, | |
| "learning_rate": 5.813008130081301e-06, | |
| "loss": 1.7755, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.4268292682926829, | |
| "grad_norm": 0.3990177810192108, | |
| "learning_rate": 5.772357723577237e-06, | |
| "loss": 1.7391, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.43089430894308944, | |
| "grad_norm": 0.4822699725627899, | |
| "learning_rate": 5.731707317073171e-06, | |
| "loss": 1.7601, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4349593495934959, | |
| "grad_norm": 0.464248925447464, | |
| "learning_rate": 5.691056910569106e-06, | |
| "loss": 1.7951, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.43902439024390244, | |
| "grad_norm": 0.432405948638916, | |
| "learning_rate": 5.650406504065041e-06, | |
| "loss": 1.7501, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.44308943089430897, | |
| "grad_norm": 0.4115266501903534, | |
| "learning_rate": 5.609756097560977e-06, | |
| "loss": 1.8129, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.44715447154471544, | |
| "grad_norm": 0.39715027809143066, | |
| "learning_rate": 5.569105691056911e-06, | |
| "loss": 1.736, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.45121951219512196, | |
| "grad_norm": 0.4466915726661682, | |
| "learning_rate": 5.528455284552846e-06, | |
| "loss": 1.7885, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.45528455284552843, | |
| "grad_norm": 0.42750024795532227, | |
| "learning_rate": 5.487804878048781e-06, | |
| "loss": 1.7642, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.45934959349593496, | |
| "grad_norm": 0.41643035411834717, | |
| "learning_rate": 5.447154471544716e-06, | |
| "loss": 1.7121, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.4634146341463415, | |
| "grad_norm": 0.4076242744922638, | |
| "learning_rate": 5.4065040650406504e-06, | |
| "loss": 1.7574, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.46747967479674796, | |
| "grad_norm": 0.45081648230552673, | |
| "learning_rate": 5.365853658536586e-06, | |
| "loss": 1.7505, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.4715447154471545, | |
| "grad_norm": 0.4501321017742157, | |
| "learning_rate": 5.32520325203252e-06, | |
| "loss": 1.7157, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.47560975609756095, | |
| "grad_norm": 0.4324929714202881, | |
| "learning_rate": 5.2845528455284555e-06, | |
| "loss": 1.763, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4796747967479675, | |
| "grad_norm": 0.49811163544654846, | |
| "learning_rate": 5.243902439024391e-06, | |
| "loss": 1.7487, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.483739837398374, | |
| "grad_norm": 0.43030911684036255, | |
| "learning_rate": 5.203252032520326e-06, | |
| "loss": 1.7058, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.4878048780487805, | |
| "grad_norm": 0.41700655221939087, | |
| "learning_rate": 5.162601626016261e-06, | |
| "loss": 1.7244, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.491869918699187, | |
| "grad_norm": 0.4175284504890442, | |
| "learning_rate": 5.121951219512195e-06, | |
| "loss": 1.7248, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.4959349593495935, | |
| "grad_norm": 0.4269621670246124, | |
| "learning_rate": 5.081300813008131e-06, | |
| "loss": 1.6854, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.4624629616737366, | |
| "learning_rate": 5.040650406504065e-06, | |
| "loss": 1.7449, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5040650406504065, | |
| "grad_norm": 0.46035370230674744, | |
| "learning_rate": 5e-06, | |
| "loss": 1.6939, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.508130081300813, | |
| "grad_norm": 0.4211556017398834, | |
| "learning_rate": 4.959349593495935e-06, | |
| "loss": 1.6919, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5121951219512195, | |
| "grad_norm": 0.40321481227874756, | |
| "learning_rate": 4.918699186991871e-06, | |
| "loss": 1.727, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.516260162601626, | |
| "grad_norm": 0.45469698309898376, | |
| "learning_rate": 4.8780487804878055e-06, | |
| "loss": 1.6792, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5203252032520326, | |
| "grad_norm": 0.4928475022315979, | |
| "learning_rate": 4.83739837398374e-06, | |
| "loss": 1.713, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.524390243902439, | |
| "grad_norm": 0.3949720561504364, | |
| "learning_rate": 4.796747967479675e-06, | |
| "loss": 1.7213, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5284552845528455, | |
| "grad_norm": 0.6022332906723022, | |
| "learning_rate": 4.75609756097561e-06, | |
| "loss": 1.6646, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.532520325203252, | |
| "grad_norm": 0.4028625786304474, | |
| "learning_rate": 4.715447154471545e-06, | |
| "loss": 1.6939, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5365853658536586, | |
| "grad_norm": 0.43009862303733826, | |
| "learning_rate": 4.67479674796748e-06, | |
| "loss": 1.6919, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.540650406504065, | |
| "grad_norm": 0.4282652735710144, | |
| "learning_rate": 4.634146341463416e-06, | |
| "loss": 1.7312, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5447154471544715, | |
| "grad_norm": 0.4238031506538391, | |
| "learning_rate": 4.59349593495935e-06, | |
| "loss": 1.6688, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5487804878048781, | |
| "grad_norm": 0.474398136138916, | |
| "learning_rate": 4.552845528455285e-06, | |
| "loss": 1.7131, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5528455284552846, | |
| "grad_norm": 0.4145808219909668, | |
| "learning_rate": 4.51219512195122e-06, | |
| "loss": 1.7134, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.556910569105691, | |
| "grad_norm": 0.5533434748649597, | |
| "learning_rate": 4.471544715447155e-06, | |
| "loss": 1.7191, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5609756097560976, | |
| "grad_norm": 0.4589686095714569, | |
| "learning_rate": 4.43089430894309e-06, | |
| "loss": 1.6795, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5650406504065041, | |
| "grad_norm": 0.42821556329727173, | |
| "learning_rate": 4.390243902439025e-06, | |
| "loss": 1.7132, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5691056910569106, | |
| "grad_norm": 0.5176672339439392, | |
| "learning_rate": 4.34959349593496e-06, | |
| "loss": 1.6831, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.573170731707317, | |
| "grad_norm": 0.45410770177841187, | |
| "learning_rate": 4.308943089430894e-06, | |
| "loss": 1.6613, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5772357723577236, | |
| "grad_norm": 0.4067547023296356, | |
| "learning_rate": 4.268292682926829e-06, | |
| "loss": 1.706, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.5813008130081301, | |
| "grad_norm": 0.4407687187194824, | |
| "learning_rate": 4.227642276422765e-06, | |
| "loss": 1.7168, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.5853658536585366, | |
| "grad_norm": 0.4558040201663971, | |
| "learning_rate": 4.1869918699186995e-06, | |
| "loss": 1.6783, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5894308943089431, | |
| "grad_norm": 0.47496524453163147, | |
| "learning_rate": 4.146341463414634e-06, | |
| "loss": 1.6927, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5934959349593496, | |
| "grad_norm": 0.4712445139884949, | |
| "learning_rate": 4.10569105691057e-06, | |
| "loss": 1.7105, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5975609756097561, | |
| "grad_norm": 0.527227520942688, | |
| "learning_rate": 4.0650406504065046e-06, | |
| "loss": 1.7476, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6016260162601627, | |
| "grad_norm": 0.5302590131759644, | |
| "learning_rate": 4.024390243902439e-06, | |
| "loss": 1.7037, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6056910569105691, | |
| "grad_norm": 0.4613666534423828, | |
| "learning_rate": 3.983739837398374e-06, | |
| "loss": 1.7034, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6097560975609756, | |
| "grad_norm": 0.43386778235435486, | |
| "learning_rate": 3.943089430894309e-06, | |
| "loss": 1.7165, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6138211382113821, | |
| "grad_norm": 0.4549228847026825, | |
| "learning_rate": 3.902439024390244e-06, | |
| "loss": 1.7262, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6178861788617886, | |
| "grad_norm": 0.4825079143047333, | |
| "learning_rate": 3.861788617886179e-06, | |
| "loss": 1.6685, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6219512195121951, | |
| "grad_norm": 0.45418936014175415, | |
| "learning_rate": 3.821138211382115e-06, | |
| "loss": 1.6906, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6260162601626016, | |
| "grad_norm": 0.47858789563179016, | |
| "learning_rate": 3.780487804878049e-06, | |
| "loss": 1.6736, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6300813008130082, | |
| "grad_norm": 0.4218636751174927, | |
| "learning_rate": 3.7398373983739838e-06, | |
| "loss": 1.6743, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6341463414634146, | |
| "grad_norm": 0.42544764280319214, | |
| "learning_rate": 3.699186991869919e-06, | |
| "loss": 1.6787, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6382113821138211, | |
| "grad_norm": 0.4026515781879425, | |
| "learning_rate": 3.6585365853658537e-06, | |
| "loss": 1.7407, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.6422764227642277, | |
| "grad_norm": 0.5130017399787903, | |
| "learning_rate": 3.6178861788617893e-06, | |
| "loss": 1.6421, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6463414634146342, | |
| "grad_norm": 0.4023183584213257, | |
| "learning_rate": 3.577235772357724e-06, | |
| "loss": 1.6691, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6504065040650406, | |
| "grad_norm": 0.4857276678085327, | |
| "learning_rate": 3.5365853658536588e-06, | |
| "loss": 1.7019, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6544715447154471, | |
| "grad_norm": 0.4183124601840973, | |
| "learning_rate": 3.495934959349594e-06, | |
| "loss": 1.6345, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6585365853658537, | |
| "grad_norm": 0.5401263236999512, | |
| "learning_rate": 3.4552845528455287e-06, | |
| "loss": 1.6823, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.6626016260162602, | |
| "grad_norm": 0.4167156219482422, | |
| "learning_rate": 3.414634146341464e-06, | |
| "loss": 1.6945, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.5526067018508911, | |
| "learning_rate": 3.3739837398373986e-06, | |
| "loss": 1.69, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.6707317073170732, | |
| "grad_norm": 0.43190276622772217, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.6771, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6747967479674797, | |
| "grad_norm": 0.48610788583755493, | |
| "learning_rate": 3.292682926829269e-06, | |
| "loss": 1.6998, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6788617886178862, | |
| "grad_norm": 0.4015091359615326, | |
| "learning_rate": 3.2520325203252037e-06, | |
| "loss": 1.6238, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.6829268292682927, | |
| "grad_norm": 0.40037909150123596, | |
| "learning_rate": 3.211382113821139e-06, | |
| "loss": 1.6885, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.6869918699186992, | |
| "grad_norm": 0.4132348299026489, | |
| "learning_rate": 3.1707317073170736e-06, | |
| "loss": 1.6955, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6910569105691057, | |
| "grad_norm": 0.5056420564651489, | |
| "learning_rate": 3.1300813008130083e-06, | |
| "loss": 1.6426, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6951219512195121, | |
| "grad_norm": 0.4648841619491577, | |
| "learning_rate": 3.0894308943089435e-06, | |
| "loss": 1.66, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6991869918699187, | |
| "grad_norm": 0.4295191168785095, | |
| "learning_rate": 3.0487804878048782e-06, | |
| "loss": 1.5901, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7032520325203252, | |
| "grad_norm": 0.42674049735069275, | |
| "learning_rate": 3.0081300813008134e-06, | |
| "loss": 1.6857, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7073170731707317, | |
| "grad_norm": 0.40749216079711914, | |
| "learning_rate": 2.967479674796748e-06, | |
| "loss": 1.6882, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7113821138211383, | |
| "grad_norm": 0.4567708969116211, | |
| "learning_rate": 2.926829268292683e-06, | |
| "loss": 1.6784, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7154471544715447, | |
| "grad_norm": 0.5232097506523132, | |
| "learning_rate": 2.8861788617886185e-06, | |
| "loss": 1.6867, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7195121951219512, | |
| "grad_norm": 0.4262886047363281, | |
| "learning_rate": 2.845528455284553e-06, | |
| "loss": 1.6745, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7235772357723578, | |
| "grad_norm": 0.4635554850101471, | |
| "learning_rate": 2.8048780487804884e-06, | |
| "loss": 1.6693, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7276422764227642, | |
| "grad_norm": 0.4350730776786804, | |
| "learning_rate": 2.764227642276423e-06, | |
| "loss": 1.6796, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7317073170731707, | |
| "grad_norm": 0.4890363812446594, | |
| "learning_rate": 2.723577235772358e-06, | |
| "loss": 1.6825, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7357723577235772, | |
| "grad_norm": 0.42314425110816956, | |
| "learning_rate": 2.682926829268293e-06, | |
| "loss": 1.654, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7398373983739838, | |
| "grad_norm": 0.4569300413131714, | |
| "learning_rate": 2.6422764227642278e-06, | |
| "loss": 1.6615, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7439024390243902, | |
| "grad_norm": 0.5300145149230957, | |
| "learning_rate": 2.601626016260163e-06, | |
| "loss": 1.6732, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.7479674796747967, | |
| "grad_norm": 0.4384273886680603, | |
| "learning_rate": 2.5609756097560977e-06, | |
| "loss": 1.6822, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7520325203252033, | |
| "grad_norm": 0.40823447704315186, | |
| "learning_rate": 2.5203252032520324e-06, | |
| "loss": 1.6947, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7560975609756098, | |
| "grad_norm": 0.44767969846725464, | |
| "learning_rate": 2.4796747967479676e-06, | |
| "loss": 1.7653, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.7601626016260162, | |
| "grad_norm": 0.42057687044143677, | |
| "learning_rate": 2.4390243902439027e-06, | |
| "loss": 1.6582, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.7642276422764228, | |
| "grad_norm": 0.6323032379150391, | |
| "learning_rate": 2.3983739837398375e-06, | |
| "loss": 1.6377, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.7682926829268293, | |
| "grad_norm": 0.46569883823394775, | |
| "learning_rate": 2.3577235772357727e-06, | |
| "loss": 1.6864, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.7723577235772358, | |
| "grad_norm": 0.4475976228713989, | |
| "learning_rate": 2.317073170731708e-06, | |
| "loss": 1.6995, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.7764227642276422, | |
| "grad_norm": 0.42551514506340027, | |
| "learning_rate": 2.2764227642276426e-06, | |
| "loss": 1.6745, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.7804878048780488, | |
| "grad_norm": 0.4114595055580139, | |
| "learning_rate": 2.2357723577235773e-06, | |
| "loss": 1.6414, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.7845528455284553, | |
| "grad_norm": 0.4425772726535797, | |
| "learning_rate": 2.1951219512195125e-06, | |
| "loss": 1.668, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7886178861788617, | |
| "grad_norm": 0.4178754985332489, | |
| "learning_rate": 2.154471544715447e-06, | |
| "loss": 1.6172, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.7926829268292683, | |
| "grad_norm": 0.526268482208252, | |
| "learning_rate": 2.1138211382113824e-06, | |
| "loss": 1.6877, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.7967479674796748, | |
| "grad_norm": 0.46816593408584595, | |
| "learning_rate": 2.073170731707317e-06, | |
| "loss": 1.6405, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.8008130081300813, | |
| "grad_norm": 0.42700088024139404, | |
| "learning_rate": 2.0325203252032523e-06, | |
| "loss": 1.6716, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8048780487804879, | |
| "grad_norm": 0.46738067269325256, | |
| "learning_rate": 1.991869918699187e-06, | |
| "loss": 1.6887, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8089430894308943, | |
| "grad_norm": 0.4532390534877777, | |
| "learning_rate": 1.951219512195122e-06, | |
| "loss": 1.6998, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8130081300813008, | |
| "grad_norm": 0.42802944779396057, | |
| "learning_rate": 1.9105691056910574e-06, | |
| "loss": 1.6087, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8170731707317073, | |
| "grad_norm": 0.547961413860321, | |
| "learning_rate": 1.8699186991869919e-06, | |
| "loss": 1.6467, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.8211382113821138, | |
| "grad_norm": 0.4223916530609131, | |
| "learning_rate": 1.8292682926829268e-06, | |
| "loss": 1.6578, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.8252032520325203, | |
| "grad_norm": 0.5189731121063232, | |
| "learning_rate": 1.788617886178862e-06, | |
| "loss": 1.7209, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.8292682926829268, | |
| "grad_norm": 0.45684292912483215, | |
| "learning_rate": 1.747967479674797e-06, | |
| "loss": 1.6436, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 0.4165087640285492, | |
| "learning_rate": 1.707317073170732e-06, | |
| "loss": 1.5974, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.8373983739837398, | |
| "grad_norm": 0.4212961196899414, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 1.6543, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.8414634146341463, | |
| "grad_norm": 0.43846744298934937, | |
| "learning_rate": 1.6260162601626018e-06, | |
| "loss": 1.6136, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.8455284552845529, | |
| "grad_norm": 0.6994388699531555, | |
| "learning_rate": 1.5853658536585368e-06, | |
| "loss": 1.7406, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.8495934959349594, | |
| "grad_norm": 0.5141953825950623, | |
| "learning_rate": 1.5447154471544717e-06, | |
| "loss": 1.6644, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.8536585365853658, | |
| "grad_norm": 0.49255314469337463, | |
| "learning_rate": 1.5040650406504067e-06, | |
| "loss": 1.6755, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.8577235772357723, | |
| "grad_norm": 0.5186207890510559, | |
| "learning_rate": 1.4634146341463414e-06, | |
| "loss": 1.5835, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.8617886178861789, | |
| "grad_norm": 0.41615086793899536, | |
| "learning_rate": 1.4227642276422766e-06, | |
| "loss": 1.6537, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.8658536585365854, | |
| "grad_norm": 0.5542620420455933, | |
| "learning_rate": 1.3821138211382116e-06, | |
| "loss": 1.642, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.8699186991869918, | |
| "grad_norm": 0.4988357424736023, | |
| "learning_rate": 1.3414634146341465e-06, | |
| "loss": 1.621, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.8739837398373984, | |
| "grad_norm": 0.3991510272026062, | |
| "learning_rate": 1.3008130081300815e-06, | |
| "loss": 1.6462, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.8780487804878049, | |
| "grad_norm": 0.3977624475955963, | |
| "learning_rate": 1.2601626016260162e-06, | |
| "loss": 1.6328, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.8821138211382114, | |
| "grad_norm": 0.4829718768596649, | |
| "learning_rate": 1.2195121951219514e-06, | |
| "loss": 1.6878, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.8861788617886179, | |
| "grad_norm": 0.5087659358978271, | |
| "learning_rate": 1.1788617886178863e-06, | |
| "loss": 1.6832, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.8902439024390244, | |
| "grad_norm": 0.4450823664665222, | |
| "learning_rate": 1.1382113821138213e-06, | |
| "loss": 1.6896, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.8943089430894309, | |
| "grad_norm": 0.5218344926834106, | |
| "learning_rate": 1.0975609756097562e-06, | |
| "loss": 1.6422, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.8983739837398373, | |
| "grad_norm": 0.43621551990509033, | |
| "learning_rate": 1.0569105691056912e-06, | |
| "loss": 1.6496, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.9024390243902439, | |
| "grad_norm": 0.5574150085449219, | |
| "learning_rate": 1.0162601626016261e-06, | |
| "loss": 1.6791, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.9065040650406504, | |
| "grad_norm": 0.47324275970458984, | |
| "learning_rate": 9.75609756097561e-07, | |
| "loss": 1.6603, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.9105691056910569, | |
| "grad_norm": 0.5460381507873535, | |
| "learning_rate": 9.349593495934959e-07, | |
| "loss": 1.6776, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.9146341463414634, | |
| "grad_norm": 0.506424069404602, | |
| "learning_rate": 8.94308943089431e-07, | |
| "loss": 1.6737, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.9186991869918699, | |
| "grad_norm": 0.48858654499053955, | |
| "learning_rate": 8.53658536585366e-07, | |
| "loss": 1.6779, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.9227642276422764, | |
| "grad_norm": 0.43115416169166565, | |
| "learning_rate": 8.130081300813009e-07, | |
| "loss": 1.6291, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.926829268292683, | |
| "grad_norm": 0.5290559530258179, | |
| "learning_rate": 7.723577235772359e-07, | |
| "loss": 1.6664, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.9308943089430894, | |
| "grad_norm": 0.3979451358318329, | |
| "learning_rate": 7.317073170731707e-07, | |
| "loss": 1.6533, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.9349593495934959, | |
| "grad_norm": 0.437351256608963, | |
| "learning_rate": 6.910569105691058e-07, | |
| "loss": 1.6444, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.9390243902439024, | |
| "grad_norm": 0.41036128997802734, | |
| "learning_rate": 6.504065040650407e-07, | |
| "loss": 1.653, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.943089430894309, | |
| "grad_norm": 0.47654396295547485, | |
| "learning_rate": 6.097560975609757e-07, | |
| "loss": 1.6934, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.9471544715447154, | |
| "grad_norm": 0.467517614364624, | |
| "learning_rate": 5.691056910569106e-07, | |
| "loss": 1.6268, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.9512195121951219, | |
| "grad_norm": 0.44289228320121765, | |
| "learning_rate": 5.284552845528456e-07, | |
| "loss": 1.5799, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.9552845528455285, | |
| "grad_norm": 0.4003511369228363, | |
| "learning_rate": 4.878048780487805e-07, | |
| "loss": 1.6795, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.959349593495935, | |
| "grad_norm": 0.39896637201309204, | |
| "learning_rate": 4.471544715447155e-07, | |
| "loss": 1.6185, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.9634146341463414, | |
| "grad_norm": 0.4294638931751251, | |
| "learning_rate": 4.0650406504065046e-07, | |
| "loss": 1.6522, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.967479674796748, | |
| "grad_norm": 0.46294713020324707, | |
| "learning_rate": 3.6585365853658536e-07, | |
| "loss": 1.6313, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.9715447154471545, | |
| "grad_norm": 0.43310508131980896, | |
| "learning_rate": 3.2520325203252037e-07, | |
| "loss": 1.6483, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.975609756097561, | |
| "grad_norm": 0.46582019329071045, | |
| "learning_rate": 2.845528455284553e-07, | |
| "loss": 1.6394, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.9796747967479674, | |
| "grad_norm": 0.44841283559799194, | |
| "learning_rate": 2.439024390243903e-07, | |
| "loss": 1.6587, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.983739837398374, | |
| "grad_norm": 0.4253198802471161, | |
| "learning_rate": 2.0325203252032523e-07, | |
| "loss": 1.6218, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.9878048780487805, | |
| "grad_norm": 0.402408242225647, | |
| "learning_rate": 1.6260162601626018e-07, | |
| "loss": 1.6548, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.991869918699187, | |
| "grad_norm": 0.3985183835029602, | |
| "learning_rate": 1.2195121951219514e-07, | |
| "loss": 1.6401, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.9959349593495935, | |
| "grad_norm": 0.5187343955039978, | |
| "learning_rate": 8.130081300813009e-08, | |
| "loss": 1.6755, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.4178217053413391, | |
| "learning_rate": 4.0650406504065046e-08, | |
| "loss": 1.5977, | |
| "step": 246 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 246, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 0, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.0971527577075712e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |