Commit ·
c848f02
1
Parent(s): d58f730
Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +253 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,253 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 18.06451612903226,
|
| 5 |
+
"global_step": 140,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.9,
|
| 12 |
+
"eval_accuracy": 0.7458100558659218,
|
| 13 |
+
"eval_f1": 0.0,
|
| 14 |
+
"eval_loss": 1.2875592708587646,
|
| 15 |
+
"eval_precision": 0.0,
|
| 16 |
+
"eval_recall": 0.0,
|
| 17 |
+
"eval_runtime": 0.0647,
|
| 18 |
+
"eval_samples_per_second": 123.718,
|
| 19 |
+
"eval_steps_per_second": 15.465,
|
| 20 |
+
"step": 7
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.94,
|
| 24 |
+
"eval_accuracy": 0.7458100558659218,
|
| 25 |
+
"eval_f1": 0.0,
|
| 26 |
+
"eval_loss": 0.9501006007194519,
|
| 27 |
+
"eval_precision": 0.0,
|
| 28 |
+
"eval_recall": 0.0,
|
| 29 |
+
"eval_runtime": 0.1181,
|
| 30 |
+
"eval_samples_per_second": 67.741,
|
| 31 |
+
"eval_steps_per_second": 8.468,
|
| 32 |
+
"step": 15
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.97,
|
| 36 |
+
"eval_accuracy": 0.7946927374301676,
|
| 37 |
+
"eval_f1": 0.1605839416058394,
|
| 38 |
+
"eval_loss": 0.7821663618087769,
|
| 39 |
+
"eval_precision": 0.15942028985507245,
|
| 40 |
+
"eval_recall": 0.16176470588235295,
|
| 41 |
+
"eval_runtime": 0.0656,
|
| 42 |
+
"eval_samples_per_second": 121.917,
|
| 43 |
+
"eval_steps_per_second": 15.24,
|
| 44 |
+
"step": 23
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.0,
|
| 48 |
+
"eval_accuracy": 0.8240223463687151,
|
| 49 |
+
"eval_f1": 0.3636363636363636,
|
| 50 |
+
"eval_loss": 0.6530368328094482,
|
| 51 |
+
"eval_precision": 0.3466666666666667,
|
| 52 |
+
"eval_recall": 0.38235294117647056,
|
| 53 |
+
"eval_runtime": 0.0584,
|
| 54 |
+
"eval_samples_per_second": 137.09,
|
| 55 |
+
"eval_steps_per_second": 17.136,
|
| 56 |
+
"step": 31
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 4.9,
|
| 60 |
+
"eval_accuracy": 0.8519553072625698,
|
| 61 |
+
"eval_f1": 0.5454545454545454,
|
| 62 |
+
"eval_loss": 0.5587103366851807,
|
| 63 |
+
"eval_precision": 0.52,
|
| 64 |
+
"eval_recall": 0.5735294117647058,
|
| 65 |
+
"eval_runtime": 0.0576,
|
| 66 |
+
"eval_samples_per_second": 138.924,
|
| 67 |
+
"eval_steps_per_second": 17.366,
|
| 68 |
+
"step": 38
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 5.94,
|
| 72 |
+
"eval_accuracy": 0.88268156424581,
|
| 73 |
+
"eval_f1": 0.5732484076433122,
|
| 74 |
+
"eval_loss": 0.501702070236206,
|
| 75 |
+
"eval_precision": 0.5056179775280899,
|
| 76 |
+
"eval_recall": 0.6617647058823529,
|
| 77 |
+
"eval_runtime": 0.0775,
|
| 78 |
+
"eval_samples_per_second": 103.223,
|
| 79 |
+
"eval_steps_per_second": 12.903,
|
| 80 |
+
"step": 46
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 6.97,
|
| 84 |
+
"eval_accuracy": 0.9078212290502793,
|
| 85 |
+
"eval_f1": 0.5875,
|
| 86 |
+
"eval_loss": 0.43670037388801575,
|
| 87 |
+
"eval_precision": 0.5108695652173914,
|
| 88 |
+
"eval_recall": 0.6911764705882353,
|
| 89 |
+
"eval_runtime": 0.094,
|
| 90 |
+
"eval_samples_per_second": 85.063,
|
| 91 |
+
"eval_steps_per_second": 10.633,
|
| 92 |
+
"step": 54
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 8.0,
|
| 96 |
+
"eval_accuracy": 0.8966480446927374,
|
| 97 |
+
"eval_f1": 0.5987261146496815,
|
| 98 |
+
"eval_loss": 0.4147437810897827,
|
| 99 |
+
"eval_precision": 0.5280898876404494,
|
| 100 |
+
"eval_recall": 0.6911764705882353,
|
| 101 |
+
"eval_runtime": 0.0593,
|
| 102 |
+
"eval_samples_per_second": 134.952,
|
| 103 |
+
"eval_steps_per_second": 16.869,
|
| 104 |
+
"step": 62
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 8.9,
|
| 108 |
+
"eval_accuracy": 0.9175977653631285,
|
| 109 |
+
"eval_f1": 0.6933333333333332,
|
| 110 |
+
"eval_loss": 0.40070846676826477,
|
| 111 |
+
"eval_precision": 0.6341463414634146,
|
| 112 |
+
"eval_recall": 0.7647058823529411,
|
| 113 |
+
"eval_runtime": 0.0605,
|
| 114 |
+
"eval_samples_per_second": 132.124,
|
| 115 |
+
"eval_steps_per_second": 16.515,
|
| 116 |
+
"step": 69
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 9.94,
|
| 120 |
+
"eval_accuracy": 0.914804469273743,
|
| 121 |
+
"eval_f1": 0.6887417218543046,
|
| 122 |
+
"eval_loss": 0.3914220929145813,
|
| 123 |
+
"eval_precision": 0.6265060240963856,
|
| 124 |
+
"eval_recall": 0.7647058823529411,
|
| 125 |
+
"eval_runtime": 0.0705,
|
| 126 |
+
"eval_samples_per_second": 113.511,
|
| 127 |
+
"eval_steps_per_second": 14.189,
|
| 128 |
+
"step": 77
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 10.97,
|
| 132 |
+
"eval_accuracy": 0.9231843575418994,
|
| 133 |
+
"eval_f1": 0.7012987012987012,
|
| 134 |
+
"eval_loss": 0.3615421652793884,
|
| 135 |
+
"eval_precision": 0.627906976744186,
|
| 136 |
+
"eval_recall": 0.7941176470588235,
|
| 137 |
+
"eval_runtime": 0.0583,
|
| 138 |
+
"eval_samples_per_second": 137.228,
|
| 139 |
+
"eval_steps_per_second": 17.153,
|
| 140 |
+
"step": 85
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 12.0,
|
| 144 |
+
"eval_accuracy": 0.9245810055865922,
|
| 145 |
+
"eval_f1": 0.7397260273972601,
|
| 146 |
+
"eval_loss": 0.3441668450832367,
|
| 147 |
+
"eval_precision": 0.6923076923076923,
|
| 148 |
+
"eval_recall": 0.7941176470588235,
|
| 149 |
+
"eval_runtime": 0.088,
|
| 150 |
+
"eval_samples_per_second": 90.913,
|
| 151 |
+
"eval_steps_per_second": 11.364,
|
| 152 |
+
"step": 93
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 12.9,
|
| 156 |
+
"eval_accuracy": 0.9287709497206704,
|
| 157 |
+
"eval_f1": 0.7586206896551725,
|
| 158 |
+
"eval_loss": 0.3498682677745819,
|
| 159 |
+
"eval_precision": 0.7142857142857143,
|
| 160 |
+
"eval_recall": 0.8088235294117647,
|
| 161 |
+
"eval_runtime": 0.059,
|
| 162 |
+
"eval_samples_per_second": 135.636,
|
| 163 |
+
"eval_steps_per_second": 16.955,
|
| 164 |
+
"step": 100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 13.94,
|
| 168 |
+
"eval_accuracy": 0.9301675977653632,
|
| 169 |
+
"eval_f1": 0.7837837837837837,
|
| 170 |
+
"eval_loss": 0.35167065262794495,
|
| 171 |
+
"eval_precision": 0.725,
|
| 172 |
+
"eval_recall": 0.8529411764705882,
|
| 173 |
+
"eval_runtime": 0.0789,
|
| 174 |
+
"eval_samples_per_second": 101.355,
|
| 175 |
+
"eval_steps_per_second": 12.669,
|
| 176 |
+
"step": 108
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 14.97,
|
| 180 |
+
"eval_accuracy": 0.9273743016759777,
|
| 181 |
+
"eval_f1": 0.7482993197278912,
|
| 182 |
+
"eval_loss": 0.35690194368362427,
|
| 183 |
+
"eval_precision": 0.6962025316455697,
|
| 184 |
+
"eval_recall": 0.8088235294117647,
|
| 185 |
+
"eval_runtime": 0.0589,
|
| 186 |
+
"eval_samples_per_second": 135.816,
|
| 187 |
+
"eval_steps_per_second": 16.977,
|
| 188 |
+
"step": 116
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 16.0,
|
| 192 |
+
"eval_accuracy": 0.9287709497206704,
|
| 193 |
+
"eval_f1": 0.761904761904762,
|
| 194 |
+
"eval_loss": 0.3552975058555603,
|
| 195 |
+
"eval_precision": 0.7088607594936709,
|
| 196 |
+
"eval_recall": 0.8235294117647058,
|
| 197 |
+
"eval_runtime": 0.0587,
|
| 198 |
+
"eval_samples_per_second": 136.349,
|
| 199 |
+
"eval_steps_per_second": 17.044,
|
| 200 |
+
"step": 124
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 16.9,
|
| 204 |
+
"eval_accuracy": 0.9259776536312849,
|
| 205 |
+
"eval_f1": 0.767123287671233,
|
| 206 |
+
"eval_loss": 0.35346120595932007,
|
| 207 |
+
"eval_precision": 0.717948717948718,
|
| 208 |
+
"eval_recall": 0.8235294117647058,
|
| 209 |
+
"eval_runtime": 0.0838,
|
| 210 |
+
"eval_samples_per_second": 95.442,
|
| 211 |
+
"eval_steps_per_second": 11.93,
|
| 212 |
+
"step": 131
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 17.94,
|
| 216 |
+
"eval_accuracy": 0.9259776536312849,
|
| 217 |
+
"eval_f1": 0.7567567567567567,
|
| 218 |
+
"eval_loss": 0.3554823398590088,
|
| 219 |
+
"eval_precision": 0.7,
|
| 220 |
+
"eval_recall": 0.8235294117647058,
|
| 221 |
+
"eval_runtime": 0.061,
|
| 222 |
+
"eval_samples_per_second": 131.23,
|
| 223 |
+
"eval_steps_per_second": 16.404,
|
| 224 |
+
"step": 139
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 18.06,
|
| 228 |
+
"eval_accuracy": 0.9259776536312849,
|
| 229 |
+
"eval_f1": 0.7567567567567567,
|
| 230 |
+
"eval_loss": 0.355307012796402,
|
| 231 |
+
"eval_precision": 0.7,
|
| 232 |
+
"eval_recall": 0.8235294117647058,
|
| 233 |
+
"eval_runtime": 0.0634,
|
| 234 |
+
"eval_samples_per_second": 126.271,
|
| 235 |
+
"eval_steps_per_second": 15.784,
|
| 236 |
+
"step": 140
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 18.06,
|
| 240 |
+
"step": 140,
|
| 241 |
+
"total_flos": 142823207204352.0,
|
| 242 |
+
"train_loss": 0.3916678019932338,
|
| 243 |
+
"train_runtime": 1023.1328,
|
| 244 |
+
"train_samples_per_second": 2.365,
|
| 245 |
+
"train_steps_per_second": 0.137
|
| 246 |
+
}
|
| 247 |
+
],
|
| 248 |
+
"max_steps": 140,
|
| 249 |
+
"num_train_epochs": 20,
|
| 250 |
+
"total_flos": 142823207204352.0,
|
| 251 |
+
"trial_name": null,
|
| 252 |
+
"trial_params": null
|
| 253 |
+
}
|