|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 5.955918367346939, |
|
"eval_steps": 50.0, |
|
"global_step": 114, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.6666666666666667e-06, |
|
"loss": 1.8797, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 3.3333333333333333e-06, |
|
"loss": 1.0534, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 5e-06, |
|
"loss": 0.8715, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 6.666666666666667e-06, |
|
"loss": 1.7966, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 8.333333333333334e-06, |
|
"loss": 3.2048, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 1e-05, |
|
"loss": 0.8377, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.1666666666666668e-05, |
|
"loss": 1.1957, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.3333333333333333e-05, |
|
"loss": 1.2774, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.5000000000000002e-05, |
|
"loss": 1.1601, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.6666666666666667e-05, |
|
"loss": 1.1541, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 1.8333333333333333e-05, |
|
"loss": 1.314, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 2e-05, |
|
"loss": 1.7869, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 1.999525719713366e-05, |
|
"loss": 1.9069, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 1.9981033287370443e-05, |
|
"loss": 1.9247, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 1.9957341762950346e-05, |
|
"loss": 3.4828, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 1.992420509671936e-05, |
|
"loss": 0.7285, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 1.9881654720812594e-05, |
|
"loss": 0.962, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.982973099683902e-05, |
|
"loss": 1.6071, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 1.9768483177596008e-05, |
|
"loss": 3.0727, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 1.9697969360350098e-05, |
|
"loss": 2.4161, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 1.961825643172819e-05, |
|
"loss": 1.0778, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 1.9529420004271568e-05, |
|
"loss": 0.8836, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 1.9431544344712776e-05, |
|
"loss": 1.7842, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 1.932472229404356e-05, |
|
"loss": 2.3738, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 1.920905517944954e-05, |
|
"loss": 0.8902, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 1.9084652718195237e-05, |
|
"loss": 1.3527, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 1.8951632913550625e-05, |
|
"loss": 1.3758, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 1.8810121942857848e-05, |
|
"loss": 0.9498, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 1.866025403784439e-05, |
|
"loss": 1.2479, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 1.8502171357296144e-05, |
|
"loss": 1.0556, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 1.8336023852211197e-05, |
|
"loss": 1.1853, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 1.816196912356222e-05, |
|
"loss": 2.0907, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 1.7980172272802398e-05, |
|
"loss": 0.8613, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 1.7790805745256703e-05, |
|
"loss": 3.3203, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 1.7594049166547073e-05, |
|
"loss": 0.6468, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 1.7390089172206594e-05, |
|
"loss": 0.8549, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 1.717911923064442e-05, |
|
"loss": 1.2893, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 1.696133945962927e-05, |
|
"loss": 2.2485, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 1.6736956436465573e-05, |
|
"loss": 2.4505, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 1.650618300204242e-05, |
|
"loss": 0.616, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 1.626923805894107e-05, |
|
"loss": 0.6972, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.6026346363792565e-05, |
|
"loss": 1.3938, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 2.25, |
|
"learning_rate": 1.5777738314082514e-05, |
|
"loss": 1.9997, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 2.3, |
|
"learning_rate": 1.552364972960506e-05, |
|
"loss": 0.7067, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 2.35, |
|
"learning_rate": 1.526432162877356e-05, |
|
"loss": 0.9563, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 2.4, |
|
"learning_rate": 1.5000000000000002e-05, |
|
"loss": 0.8421, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 1.4730935568360103e-05, |
|
"loss": 0.6379, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 1.4457383557765385e-05, |
|
"loss": 1.3021, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 1.4179603448867836e-05, |
|
"loss": 0.4532, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 2.61, |
|
"learning_rate": 1.3897858732926794e-05, |
|
"loss": 1.0642, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 2.66, |
|
"learning_rate": 1.3612416661871532e-05, |
|
"loss": 1.7012, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 2.72, |
|
"learning_rate": 1.3323547994796597e-05, |
|
"loss": 0.4655, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 2.77, |
|
"learning_rate": 1.3031526741130435e-05, |
|
"loss": 1.4864, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 2.82, |
|
"learning_rate": 1.2736629900720832e-05, |
|
"loss": 0.4871, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 2.87, |
|
"learning_rate": 1.2439137201083772e-05, |
|
"loss": 0.6007, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 2.93, |
|
"learning_rate": 1.2139330832064975e-05, |
|
"loss": 0.7067, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 2.98, |
|
"learning_rate": 1.1837495178165706e-05, |
|
"loss": 2.339, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 3.03, |
|
"learning_rate": 1.1533916548786856e-05, |
|
"loss": 1.5665, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 3.08, |
|
"learning_rate": 1.1228882906647142e-05, |
|
"loss": 0.3888, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 3.13, |
|
"learning_rate": 1.092268359463302e-05, |
|
"loss": 0.4401, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 3.19, |
|
"learning_rate": 1.0615609061339431e-05, |
|
"loss": 0.6339, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 3.24, |
|
"learning_rate": 1.0307950585561705e-05, |
|
"loss": 1.1265, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 3.29, |
|
"learning_rate": 1e-05, |
|
"loss": 0.6132, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 3.34, |
|
"learning_rate": 9.692049414438298e-06, |
|
"loss": 0.5684, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 3.4, |
|
"learning_rate": 9.384390938660572e-06, |
|
"loss": 0.6286, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 3.45, |
|
"learning_rate": 9.07731640536698e-06, |
|
"loss": 0.3793, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 3.5, |
|
"learning_rate": 8.771117093352861e-06, |
|
"loss": 0.6226, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 3.55, |
|
"learning_rate": 8.466083451213145e-06, |
|
"loss": 0.3187, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 3.6, |
|
"learning_rate": 8.162504821834296e-06, |
|
"loss": 0.6939, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 3.66, |
|
"learning_rate": 7.860669167935028e-06, |
|
"loss": 0.9014, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 3.71, |
|
"learning_rate": 7.560862798916229e-06, |
|
"loss": 0.3323, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 3.76, |
|
"learning_rate": 7.263370099279173e-06, |
|
"loss": 0.759, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 3.81, |
|
"learning_rate": 6.968473258869566e-06, |
|
"loss": 0.4101, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 3.87, |
|
"learning_rate": 6.6764520052034054e-06, |
|
"loss": 0.3119, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 3.92, |
|
"learning_rate": 6.387583338128471e-06, |
|
"loss": 0.3196, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 3.97, |
|
"learning_rate": 6.102141267073207e-06, |
|
"loss": 1.207, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 4.02, |
|
"learning_rate": 5.82039655113217e-06, |
|
"loss": 1.1213, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 4.08, |
|
"learning_rate": 5.542616442234618e-06, |
|
"loss": 0.2656, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 4.13, |
|
"learning_rate": 5.269064431639901e-06, |
|
"loss": 0.2835, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 4.18, |
|
"learning_rate": 5.000000000000003e-06, |
|
"loss": 0.4549, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 4.23, |
|
"learning_rate": 4.7356783712264405e-06, |
|
"loss": 1.0032, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 4.28, |
|
"learning_rate": 4.476350270394942e-06, |
|
"loss": 0.4969, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 4.34, |
|
"learning_rate": 4.222261685917489e-06, |
|
"loss": 0.4467, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 4.39, |
|
"learning_rate": 3.973653636207437e-06, |
|
"loss": 0.4484, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 4.44, |
|
"learning_rate": 3.730761941058938e-06, |
|
"loss": 0.2783, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 4.49, |
|
"learning_rate": 3.493816997957582e-06, |
|
"loss": 0.5231, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 4.55, |
|
"learning_rate": 3.2630435635344283e-06, |
|
"loss": 0.2802, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 4.6, |
|
"learning_rate": 3.0386605403707347e-06, |
|
"loss": 0.4629, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 4.65, |
|
"learning_rate": 2.820880769355582e-06, |
|
"loss": 0.6277, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 4.7, |
|
"learning_rate": 2.6099108277934105e-06, |
|
"loss": 0.1864, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 4.75, |
|
"learning_rate": 2.405950833452928e-06, |
|
"loss": 0.725, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 4.81, |
|
"learning_rate": 2.209194254743295e-06, |
|
"loss": 0.3775, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 4.86, |
|
"learning_rate": 2.019827727197605e-06, |
|
"loss": 0.2099, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 4.91, |
|
"learning_rate": 1.8380308764377841e-06, |
|
"loss": 0.2252, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 4.96, |
|
"learning_rate": 1.663976147788806e-06, |
|
"loss": 1.1326, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 5.02, |
|
"learning_rate": 1.4978286427038602e-06, |
|
"loss": 0.3892, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 5.07, |
|
"learning_rate": 1.339745962155613e-06, |
|
"loss": 0.2347, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 5.12, |
|
"learning_rate": 1.1898780571421554e-06, |
|
"loss": 0.2129, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 5.17, |
|
"learning_rate": 1.0483670864493777e-06, |
|
"loss": 0.2514, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 5.22, |
|
"learning_rate": 9.153472818047627e-07, |
|
"loss": 0.4505, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 5.28, |
|
"learning_rate": 7.909448205504633e-07, |
|
"loss": 0.363, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 5.33, |
|
"learning_rate": 6.752777059564431e-07, |
|
"loss": 0.3963, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 5.38, |
|
"learning_rate": 5.684556552872256e-07, |
|
"loss": 0.289, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 5.43, |
|
"learning_rate": 4.7057999572843516e-07, |
|
"loss": 0.2103, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 5.49, |
|
"learning_rate": 3.817435682718096e-07, |
|
"loss": 0.3918, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 5.54, |
|
"learning_rate": 3.020306396499062e-07, |
|
"loss": 0.2379, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 5.59, |
|
"learning_rate": 2.315168224039932e-07, |
|
"loss": 0.4024, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 5.64, |
|
"learning_rate": 1.7026900316098217e-07, |
|
"loss": 0.612, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 5.69, |
|
"learning_rate": 1.1834527918740624e-07, |
|
"loss": 0.1498, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 5.75, |
|
"learning_rate": 7.579490328064265e-08, |
|
"loss": 0.3277, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 5.8, |
|
"learning_rate": 4.2658237049655325e-08, |
|
"loss": 0.4515, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 5.85, |
|
"learning_rate": 1.896671262955896e-08, |
|
"loss": 0.1789, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 5.9, |
|
"learning_rate": 4.74280286634099e-09, |
|
"loss": 0.156, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 5.96, |
|
"learning_rate": 0.0, |
|
"loss": 0.8604, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 5.96, |
|
"step": 114, |
|
"total_flos": 5.099950184243134e+18, |
|
"train_loss": 0.9505624028674343, |
|
"train_runtime": 9481.25, |
|
"train_samples_per_second": 6.202, |
|
"train_steps_per_second": 0.012 |
|
} |
|
], |
|
"logging_steps": 1.0, |
|
"max_steps": 114, |
|
"num_train_epochs": 6, |
|
"save_steps": 100.0, |
|
"total_flos": 5.099950184243134e+18, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|