Commit
•
61eed8a
1
Parent(s):
82e12d4
Delete trainer_state.json
Browse files- trainer_state.json +0 -156
trainer_state.json
DELETED
|
@@ -1,156 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 1.9486271036315324,
|
| 5 |
-
"global_step": 11000,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.09,
|
| 12 |
-
"learning_rate": 7.374631268436578e-05,
|
| 13 |
-
"loss": 3.876,
|
| 14 |
-
"step": 500
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.18,
|
| 18 |
-
"learning_rate": 9.696569920844328e-05,
|
| 19 |
-
"loss": 1.5925,
|
| 20 |
-
"step": 1000
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.27,
|
| 24 |
-
"learning_rate": 9.2254052016585e-05,
|
| 25 |
-
"loss": 1.2572,
|
| 26 |
-
"step": 1500
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.35,
|
| 30 |
-
"learning_rate": 8.754240482472673e-05,
|
| 31 |
-
"loss": 1.1371,
|
| 32 |
-
"step": 2000
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.44,
|
| 36 |
-
"learning_rate": 8.283075763286845e-05,
|
| 37 |
-
"loss": 1.0932,
|
| 38 |
-
"step": 2500
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.53,
|
| 42 |
-
"learning_rate": 7.811911044101018e-05,
|
| 43 |
-
"loss": 1.0323,
|
| 44 |
-
"step": 3000
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.62,
|
| 48 |
-
"learning_rate": 7.340746324915191e-05,
|
| 49 |
-
"loss": 1.0333,
|
| 50 |
-
"step": 3500
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.71,
|
| 54 |
-
"learning_rate": 6.869581605729364e-05,
|
| 55 |
-
"loss": 0.9684,
|
| 56 |
-
"step": 4000
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.8,
|
| 60 |
-
"learning_rate": 6.398416886543535e-05,
|
| 61 |
-
"loss": 0.9606,
|
| 62 |
-
"step": 4500
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.89,
|
| 66 |
-
"learning_rate": 5.927252167357709e-05,
|
| 67 |
-
"loss": 0.9397,
|
| 68 |
-
"step": 5000
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.97,
|
| 72 |
-
"learning_rate": 5.4560874481718815e-05,
|
| 73 |
-
"loss": 0.8899,
|
| 74 |
-
"step": 5500
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 1.0,
|
| 78 |
-
"eval_loss": 1.0070745944976807,
|
| 79 |
-
"eval_runtime": 18.4896,
|
| 80 |
-
"eval_samples_per_second": 176.37,
|
| 81 |
-
"eval_steps_per_second": 7.355,
|
| 82 |
-
"step": 5645
|
| 83 |
-
},
|
| 84 |
-
{
|
| 85 |
-
"epoch": 1.06,
|
| 86 |
-
"learning_rate": 4.984922728986054e-05,
|
| 87 |
-
"loss": 0.733,
|
| 88 |
-
"step": 6000
|
| 89 |
-
},
|
| 90 |
-
{
|
| 91 |
-
"epoch": 1.15,
|
| 92 |
-
"learning_rate": 4.5137580098002266e-05,
|
| 93 |
-
"loss": 0.6544,
|
| 94 |
-
"step": 6500
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 1.24,
|
| 98 |
-
"learning_rate": 4.0425932906143985e-05,
|
| 99 |
-
"loss": 0.6608,
|
| 100 |
-
"step": 7000
|
| 101 |
-
},
|
| 102 |
-
{
|
| 103 |
-
"epoch": 1.33,
|
| 104 |
-
"learning_rate": 3.571428571428572e-05,
|
| 105 |
-
"loss": 0.6493,
|
| 106 |
-
"step": 7500
|
| 107 |
-
},
|
| 108 |
-
{
|
| 109 |
-
"epoch": 1.42,
|
| 110 |
-
"learning_rate": 3.100263852242744e-05,
|
| 111 |
-
"loss": 0.6352,
|
| 112 |
-
"step": 8000
|
| 113 |
-
},
|
| 114 |
-
{
|
| 115 |
-
"epoch": 1.51,
|
| 116 |
-
"learning_rate": 2.629099133056917e-05,
|
| 117 |
-
"loss": 0.6268,
|
| 118 |
-
"step": 8500
|
| 119 |
-
},
|
| 120 |
-
{
|
| 121 |
-
"epoch": 1.59,
|
| 122 |
-
"learning_rate": 2.1579344138710894e-05,
|
| 123 |
-
"loss": 0.6172,
|
| 124 |
-
"step": 9000
|
| 125 |
-
},
|
| 126 |
-
{
|
| 127 |
-
"epoch": 1.68,
|
| 128 |
-
"learning_rate": 1.686769694685262e-05,
|
| 129 |
-
"loss": 0.6159,
|
| 130 |
-
"step": 9500
|
| 131 |
-
},
|
| 132 |
-
{
|
| 133 |
-
"epoch": 1.77,
|
| 134 |
-
"learning_rate": 1.2156049754994346e-05,
|
| 135 |
-
"loss": 0.5908,
|
| 136 |
-
"step": 10000
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 1.86,
|
| 140 |
-
"learning_rate": 7.444402563136073e-06,
|
| 141 |
-
"loss": 0.59,
|
| 142 |
-
"step": 10500
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 1.95,
|
| 146 |
-
"learning_rate": 2.7327553712777988e-06,
|
| 147 |
-
"loss": 0.582,
|
| 148 |
-
"step": 11000
|
| 149 |
-
}
|
| 150 |
-
],
|
| 151 |
-
"max_steps": 11290,
|
| 152 |
-
"num_train_epochs": 2,
|
| 153 |
-
"total_flos": 6.897685654641254e+16,
|
| 154 |
-
"trial_name": null,
|
| 155 |
-
"trial_params": null
|
| 156 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|