| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 11.034906118012294, | |
| "learning_rate": 2e-07, | |
| "loss": 3.3072, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 8.342029333286245, | |
| "learning_rate": 4e-07, | |
| "loss": 2.5181, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 11.242982411667311, | |
| "learning_rate": 6e-07, | |
| "loss": 3.091, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 7.583970040249307, | |
| "learning_rate": 8e-07, | |
| "loss": 2.0275, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 9.459350774016439, | |
| "learning_rate": 1e-06, | |
| "loss": 2.7841, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 10.513961571396786, | |
| "learning_rate": 1.2e-06, | |
| "loss": 3.1251, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 8.543813815295838, | |
| "learning_rate": 1.4e-06, | |
| "loss": 2.7579, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 6.595095020234342, | |
| "learning_rate": 1.6e-06, | |
| "loss": 2.7238, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 5.695517534101955, | |
| "learning_rate": 1.8e-06, | |
| "loss": 2.8206, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 5.393498293102003, | |
| "learning_rate": 2e-06, | |
| "loss": 2.5533, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 7.260411320337986, | |
| "learning_rate": 1.9999675557165277e-06, | |
| "loss": 2.4632, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 11.350295546296435, | |
| "learning_rate": 1.9998702249713745e-06, | |
| "loss": 3.2465, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 7.22829832667976, | |
| "learning_rate": 1.999708014080193e-06, | |
| "loss": 2.7724, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 5.264505272794518, | |
| "learning_rate": 1.9994809335686148e-06, | |
| "loss": 1.4729, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 5.487224091906534, | |
| "learning_rate": 1.9991889981715695e-06, | |
| "loss": 2.2005, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 5.639447521961718, | |
| "learning_rate": 1.9988322268323264e-06, | |
| "loss": 2.7849, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 4.877140297568701, | |
| "learning_rate": 1.9984106427012667e-06, | |
| "loss": 3.1055, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 3.7730374876185957, | |
| "learning_rate": 1.9979242731343803e-06, | |
| "loss": 2.497, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 3.657636943178752, | |
| "learning_rate": 1.997373149691491e-06, | |
| "loss": 2.9273, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 3.0729887854058964, | |
| "learning_rate": 1.99675730813421e-06, | |
| "loss": 2.5135, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 4.670717031408749, | |
| "learning_rate": 1.996076788423613e-06, | |
| "loss": 2.4479, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 2.6910831327416003, | |
| "learning_rate": 1.9953316347176486e-06, | |
| "loss": 2.0386, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 3.699623336884353, | |
| "learning_rate": 1.994521895368273e-06, | |
| "loss": 2.8066, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 3.48403675028581, | |
| "learning_rate": 1.993647622918313e-06, | |
| "loss": 2.4701, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 3.099491388558407, | |
| "learning_rate": 1.9927088740980536e-06, | |
| "loss": 2.6852, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 2.951499898355411, | |
| "learning_rate": 1.991705709821562e-06, | |
| "loss": 2.4056, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 2.7261385470933357, | |
| "learning_rate": 1.990638195182729e-06, | |
| "loss": 3.1598, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 2.729816445627831, | |
| "learning_rate": 1.989506399451051e-06, | |
| "loss": 2.5593, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 6.3601179558914005, | |
| "learning_rate": 1.9883103960671304e-06, | |
| "loss": 2.3769, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 3.3425367150604486, | |
| "learning_rate": 1.9870502626379126e-06, | |
| "loss": 2.9407, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 3.167931438919616, | |
| "learning_rate": 1.9857260809316507e-06, | |
| "loss": 2.7142, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 2.4345245228046997, | |
| "learning_rate": 1.984337936872598e-06, | |
| "loss": 2.4862, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 2.301046482260705, | |
| "learning_rate": 1.9828859205354322e-06, | |
| "loss": 3.0665, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 2.8107607204332408, | |
| "learning_rate": 1.981370126139413e-06, | |
| "loss": 3.5054, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 2.6629979731983475, | |
| "learning_rate": 1.9797906520422674e-06, | |
| "loss": 2.7669, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 2.5923595481221224, | |
| "learning_rate": 1.9781476007338054e-06, | |
| "loss": 2.9641, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 2.8444306805466013, | |
| "learning_rate": 1.976441078829272e-06, | |
| "loss": 1.8331, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 3.5565670829561022, | |
| "learning_rate": 1.974671197062428e-06, | |
| "loss": 2.701, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 3.5201013445124274, | |
| "learning_rate": 1.972838070278364e-06, | |
| "loss": 2.3367, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 2.2569984535941283, | |
| "learning_rate": 1.970941817426052e-06, | |
| "loss": 2.0169, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 2.7463871353146065, | |
| "learning_rate": 1.9689825615506204e-06, | |
| "loss": 2.0741, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 2.4202758569232974, | |
| "learning_rate": 1.9669604297853765e-06, | |
| "loss": 2.512, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 2.3710708099074864, | |
| "learning_rate": 1.9648755533435517e-06, | |
| "loss": 1.5202, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 2.918111295902255, | |
| "learning_rate": 1.9627280675097907e-06, | |
| "loss": 3.0566, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 3.2329590759683264, | |
| "learning_rate": 1.9605181116313723e-06, | |
| "loss": 2.2885, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 3.5631175201185465, | |
| "learning_rate": 1.9582458291091663e-06, | |
| "loss": 1.5685, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 2.325773119529427, | |
| "learning_rate": 1.955911367388329e-06, | |
| "loss": 2.0483, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 3.5390088616305615, | |
| "learning_rate": 1.9535148779487363e-06, | |
| "loss": 2.3026, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 3.471780417925277, | |
| "learning_rate": 1.9510565162951534e-06, | |
| "loss": 2.9149, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 3.290534820938375, | |
| "learning_rate": 1.9485364419471455e-06, | |
| "loss": 2.5526, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 2.2790349959484844, | |
| "learning_rate": 1.945954818428725e-06, | |
| "loss": 2.12, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 2.937267527964616, | |
| "learning_rate": 1.943311813257743e-06, | |
| "loss": 2.506, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 3.5635099675775397, | |
| "learning_rate": 1.9406075979350173e-06, | |
| "loss": 2.7026, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 2.882350914422191, | |
| "learning_rate": 1.9378423479332045e-06, | |
| "loss": 2.3409, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 3.2384517113771034, | |
| "learning_rate": 1.9350162426854148e-06, | |
| "loss": 2.966, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 3.3327050036846506, | |
| "learning_rate": 1.9321294655735677e-06, | |
| "loss": 1.2433, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 2.573093983791755, | |
| "learning_rate": 1.929182203916493e-06, | |
| "loss": 1.5502, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 3.68389433943638, | |
| "learning_rate": 1.9261746489577764e-06, | |
| "loss": 1.3458, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 2.7282256066111046, | |
| "learning_rate": 1.9231069958533488e-06, | |
| "loss": 1.9563, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 3.1622073079116855, | |
| "learning_rate": 1.9199794436588242e-06, | |
| "loss": 2.3207, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 2.894897190825691, | |
| "learning_rate": 1.9167921953165826e-06, | |
| "loss": 2.6743, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 2.9685942776074956, | |
| "learning_rate": 1.9135454576426007e-06, | |
| "loss": 3.2693, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 2.9212586978415422, | |
| "learning_rate": 1.9102394413130343e-06, | |
| "loss": 2.5456, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 2.535902364146159, | |
| "learning_rate": 1.9068743608505452e-06, | |
| "loss": 2.387, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 3.217651704303671, | |
| "learning_rate": 1.9034504346103822e-06, | |
| "loss": 3.1621, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 2.6086959528633287, | |
| "learning_rate": 1.899967884766212e-06, | |
| "loss": 2.2165, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 3.8950659723726524, | |
| "learning_rate": 1.8964269372957036e-06, | |
| "loss": 2.049, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 3.472843103230003, | |
| "learning_rate": 1.892827821965864e-06, | |
| "loss": 2.1215, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 5.483803674424827, | |
| "learning_rate": 1.8891707723181292e-06, | |
| "loss": 2.8693, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 2.946591324147399, | |
| "learning_rate": 1.8854560256532098e-06, | |
| "loss": 2.7123, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 3.807522896392873, | |
| "learning_rate": 1.881683823015694e-06, | |
| "loss": 2.8497, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 3.6024822728403576, | |
| "learning_rate": 1.8778544091784046e-06, | |
| "loss": 2.6366, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 3.1804694638843825, | |
| "learning_rate": 1.873968032626518e-06, | |
| "loss": 2.4264, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 2.202324783789416, | |
| "learning_rate": 1.8700249455414392e-06, | |
| "loss": 1.9495, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 2.718012853530589, | |
| "learning_rate": 1.8660254037844386e-06, | |
| "loss": 2.4048, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 2.7344487811181324, | |
| "learning_rate": 1.861969666880049e-06, | |
| "loss": 2.9311, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 2.426119637752445, | |
| "learning_rate": 1.8578579979992264e-06, | |
| "loss": 3.2567, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 2.424553353557125, | |
| "learning_rate": 1.8536906639422723e-06, | |
| "loss": 1.3894, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 2.6332246429203585, | |
| "learning_rate": 1.849467935121521e-06, | |
| "loss": 2.0805, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 2.963679112021814, | |
| "learning_rate": 1.8451900855437948e-06, | |
| "loss": 2.2385, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 3.326048125575145, | |
| "learning_rate": 1.8408573927926221e-06, | |
| "loss": 2.2506, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 2.370657089874872, | |
| "learning_rate": 1.8364701380102264e-06, | |
| "loss": 1.3311, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 2.9535264322406065, | |
| "learning_rate": 1.8320286058792843e-06, | |
| "loss": 2.2799, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 2.8249889576883387, | |
| "learning_rate": 1.82753308460445e-06, | |
| "loss": 2.8564, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 3.7437058642564813, | |
| "learning_rate": 1.8229838658936564e-06, | |
| "loss": 2.4897, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 3.2803564250465915, | |
| "learning_rate": 1.818381244939187e-06, | |
| "loss": 2.4974, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 3.2484261848308376, | |
| "learning_rate": 1.8137255203985196e-06, | |
| "loss": 2.2902, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 2.6795014358036204, | |
| "learning_rate": 1.8090169943749474e-06, | |
| "loss": 2.1316, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 2.6292578695567417, | |
| "learning_rate": 1.8042559723979767e-06, | |
| "loss": 1.9798, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 2.1422464465838176, | |
| "learning_rate": 1.7994427634035012e-06, | |
| "loss": 2.3323, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 2.993979550191501, | |
| "learning_rate": 1.7945776797137542e-06, | |
| "loss": 2.1355, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 2.5678717134911375, | |
| "learning_rate": 1.7896610370170448e-06, | |
| "loss": 1.7538, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 2.723564728060478, | |
| "learning_rate": 1.7846931543472718e-06, | |
| "loss": 2.4138, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 2.8241408405850206, | |
| "learning_rate": 1.7796743540632221e-06, | |
| "loss": 2.8358, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 2.598460561460152, | |
| "learning_rate": 1.7746049618276543e-06, | |
| "loss": 2.2555, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 3.5370024371672346, | |
| "learning_rate": 1.7694853065861659e-06, | |
| "loss": 3.0556, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 2.144317687276328, | |
| "learning_rate": 1.7643157205458482e-06, | |
| "loss": 1.4138, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 2.5108458672744893, | |
| "learning_rate": 1.7590965391537313e-06, | |
| "loss": 2.1104, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 2.9980770618295, | |
| "learning_rate": 1.7538281010750166e-06, | |
| "loss": 2.6966, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 3.166782870443781, | |
| "learning_rate": 1.748510748171101e-06, | |
| "loss": 2.1128, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 400, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8005398036480.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |