|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 9.0, |
|
"eval_steps": 500, |
|
"global_step": 81, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 2.2222222222222223e-05, |
|
"loss": 0.977, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 4.4444444444444447e-05, |
|
"loss": 0.9399, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 6.666666666666667e-05, |
|
"loss": 0.9147, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 8.888888888888889e-05, |
|
"loss": 0.9426, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 0.00011111111111111112, |
|
"loss": 0.9001, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 0.00013333333333333334, |
|
"loss": 0.8425, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 0.00015555555555555556, |
|
"loss": 0.77, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 0.00017777777777777779, |
|
"loss": 0.7601, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 0.0002, |
|
"loss": 0.6534, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 0.00019722222222222225, |
|
"loss": 0.5944, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 0.00019444444444444446, |
|
"loss": 0.5356, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 0.00019166666666666667, |
|
"loss": 0.5692, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 0.00018888888888888888, |
|
"loss": 0.5362, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 0.00018611111111111112, |
|
"loss": 0.5117, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 0.00018333333333333334, |
|
"loss": 0.5097, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 0.00018055555555555557, |
|
"loss": 0.505, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 0.00017777777777777779, |
|
"loss": 0.4853, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 0.000175, |
|
"loss": 0.4904, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 0.00017222222222222224, |
|
"loss": 0.4615, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 0.00016944444444444445, |
|
"loss": 0.4441, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 2.33, |
|
"learning_rate": 0.0001666666666666667, |
|
"loss": 0.417, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 0.0001638888888888889, |
|
"loss": 0.4511, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 0.0001611111111111111, |
|
"loss": 0.4555, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 2.67, |
|
"learning_rate": 0.00015833333333333332, |
|
"loss": 0.4463, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 2.78, |
|
"learning_rate": 0.00015555555555555556, |
|
"loss": 0.4365, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 2.89, |
|
"learning_rate": 0.00015277777777777777, |
|
"loss": 0.4109, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 0.00015000000000000001, |
|
"loss": 0.4286, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 3.11, |
|
"learning_rate": 0.00014722222222222223, |
|
"loss": 0.4158, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 3.22, |
|
"learning_rate": 0.00014444444444444444, |
|
"loss": 0.3736, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 3.33, |
|
"learning_rate": 0.00014166666666666668, |
|
"loss": 0.4037, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 3.44, |
|
"learning_rate": 0.0001388888888888889, |
|
"loss": 0.3772, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 3.56, |
|
"learning_rate": 0.00013611111111111113, |
|
"loss": 0.3651, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 3.67, |
|
"learning_rate": 0.00013333333333333334, |
|
"loss": 0.3477, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 3.78, |
|
"learning_rate": 0.00013055555555555555, |
|
"loss": 0.2898, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 3.89, |
|
"learning_rate": 0.00012777777777777776, |
|
"loss": 0.1953, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"learning_rate": 0.000125, |
|
"loss": 0.178, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 4.11, |
|
"learning_rate": 0.00012222222222222224, |
|
"loss": 0.2101, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 4.22, |
|
"learning_rate": 0.00011944444444444445, |
|
"loss": 0.2007, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 4.33, |
|
"learning_rate": 0.00011666666666666668, |
|
"loss": 0.1865, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 4.44, |
|
"learning_rate": 0.00011388888888888889, |
|
"loss": 0.1969, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 4.56, |
|
"learning_rate": 0.00011111111111111112, |
|
"loss": 0.1721, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 4.67, |
|
"learning_rate": 0.00010833333333333333, |
|
"loss": 0.1643, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 4.78, |
|
"learning_rate": 0.00010555555555555557, |
|
"loss": 0.1572, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 4.89, |
|
"learning_rate": 0.00010277777777777778, |
|
"loss": 0.1591, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"learning_rate": 0.0001, |
|
"loss": 0.1492, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 5.11, |
|
"learning_rate": 9.722222222222223e-05, |
|
"loss": 0.1498, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 5.22, |
|
"learning_rate": 9.444444444444444e-05, |
|
"loss": 0.1616, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 5.33, |
|
"learning_rate": 9.166666666666667e-05, |
|
"loss": 0.1482, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 5.44, |
|
"learning_rate": 8.888888888888889e-05, |
|
"loss": 0.1597, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 5.56, |
|
"learning_rate": 8.611111111111112e-05, |
|
"loss": 0.1424, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 5.67, |
|
"learning_rate": 8.333333333333334e-05, |
|
"loss": 0.1559, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 5.78, |
|
"learning_rate": 8.055555555555556e-05, |
|
"loss": 0.1534, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 5.89, |
|
"learning_rate": 7.777777777777778e-05, |
|
"loss": 0.1435, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"learning_rate": 7.500000000000001e-05, |
|
"loss": 0.1403, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 6.11, |
|
"learning_rate": 7.222222222222222e-05, |
|
"loss": 0.1218, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 6.22, |
|
"learning_rate": 6.944444444444444e-05, |
|
"loss": 0.1392, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 6.33, |
|
"learning_rate": 6.666666666666667e-05, |
|
"loss": 0.1482, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 6.44, |
|
"learning_rate": 6.388888888888888e-05, |
|
"loss": 0.1293, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 6.56, |
|
"learning_rate": 6.111111111111112e-05, |
|
"loss": 0.1356, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 6.67, |
|
"learning_rate": 5.833333333333334e-05, |
|
"loss": 0.163, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 6.78, |
|
"learning_rate": 5.555555555555556e-05, |
|
"loss": 0.1335, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 6.89, |
|
"learning_rate": 5.2777777777777784e-05, |
|
"loss": 0.1408, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 7.0, |
|
"learning_rate": 5e-05, |
|
"loss": 0.1422, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 7.11, |
|
"learning_rate": 4.722222222222222e-05, |
|
"loss": 0.1331, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 7.22, |
|
"learning_rate": 4.4444444444444447e-05, |
|
"loss": 0.1496, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 7.33, |
|
"learning_rate": 4.166666666666667e-05, |
|
"loss": 0.1397, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 7.44, |
|
"learning_rate": 3.888888888888889e-05, |
|
"loss": 0.1488, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 7.56, |
|
"learning_rate": 3.611111111111111e-05, |
|
"loss": 0.1112, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 7.67, |
|
"learning_rate": 3.3333333333333335e-05, |
|
"loss": 0.1117, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 7.78, |
|
"learning_rate": 3.055555555555556e-05, |
|
"loss": 0.1299, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 7.89, |
|
"learning_rate": 2.777777777777778e-05, |
|
"loss": 0.1347, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"learning_rate": 2.5e-05, |
|
"loss": 0.1364, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 8.11, |
|
"learning_rate": 2.2222222222222223e-05, |
|
"loss": 0.1064, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 8.22, |
|
"learning_rate": 1.9444444444444445e-05, |
|
"loss": 0.1367, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 8.33, |
|
"learning_rate": 1.6666666666666667e-05, |
|
"loss": 0.1491, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 8.44, |
|
"learning_rate": 1.388888888888889e-05, |
|
"loss": 0.1234, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 8.56, |
|
"learning_rate": 1.1111111111111112e-05, |
|
"loss": 0.1157, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 8.67, |
|
"learning_rate": 8.333333333333334e-06, |
|
"loss": 0.124, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 8.78, |
|
"learning_rate": 5.555555555555556e-06, |
|
"loss": 0.1321, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 8.89, |
|
"learning_rate": 2.777777777777778e-06, |
|
"loss": 0.1337, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 9.0, |
|
"learning_rate": 0.0, |
|
"loss": 0.1398, |
|
"step": 81 |
|
} |
|
], |
|
"logging_steps": 1, |
|
"max_steps": 81, |
|
"num_train_epochs": 9, |
|
"save_steps": 500, |
|
"total_flos": 1.3169647046098944e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|