./output/lstm.CEBaB_confounding.price_food_ambiance_negative.absa.5-class.seed_44
Browse files- all_results.json +16 -0
- eval_results.json +11 -0
- train_results.json +8 -0
- trainer_state.json +421 -0
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_Macro-F1": 0.7653197996917595,
|
| 4 |
+
"eval_Weighted-Macro-F1": 0.7720870459695793,
|
| 5 |
+
"eval_accuracy": 0.7705018359853121,
|
| 6 |
+
"eval_loss": 0.592179000377655,
|
| 7 |
+
"eval_runtime": 2.9703,
|
| 8 |
+
"eval_samples": 4902,
|
| 9 |
+
"eval_samples_per_second": 1650.327,
|
| 10 |
+
"eval_steps_per_second": 51.846,
|
| 11 |
+
"train_loss": 0.7294828687395368,
|
| 12 |
+
"train_runtime": 22.1174,
|
| 13 |
+
"train_samples": 4232,
|
| 14 |
+
"train_samples_per_second": 956.715,
|
| 15 |
+
"train_steps_per_second": 30.067
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_Macro-F1": 0.7653197996917595,
|
| 4 |
+
"eval_Weighted-Macro-F1": 0.7720870459695793,
|
| 5 |
+
"eval_accuracy": 0.7705018359853121,
|
| 6 |
+
"eval_loss": 0.592179000377655,
|
| 7 |
+
"eval_runtime": 2.9703,
|
| 8 |
+
"eval_samples": 4902,
|
| 9 |
+
"eval_samples_per_second": 1650.327,
|
| 10 |
+
"eval_steps_per_second": 51.846
|
| 11 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"train_loss": 0.7294828687395368,
|
| 4 |
+
"train_runtime": 22.1174,
|
| 5 |
+
"train_samples": 4232,
|
| 6 |
+
"train_samples_per_second": 956.715,
|
| 7 |
+
"train_steps_per_second": 30.067
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,421 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"global_step": 665,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.08,
|
| 12 |
+
"learning_rate": 0.0009849624060150376,
|
| 13 |
+
"loss": 1.1101,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.15,
|
| 18 |
+
"learning_rate": 0.0009699248120300752,
|
| 19 |
+
"loss": 1.0909,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.23,
|
| 24 |
+
"learning_rate": 0.0009548872180451129,
|
| 25 |
+
"loss": 1.0972,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.3,
|
| 30 |
+
"learning_rate": 0.0009398496240601504,
|
| 31 |
+
"loss": 1.0485,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.38,
|
| 36 |
+
"learning_rate": 0.000924812030075188,
|
| 37 |
+
"loss": 1.008,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.45,
|
| 42 |
+
"learning_rate": 0.0009097744360902256,
|
| 43 |
+
"loss": 1.0241,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.53,
|
| 48 |
+
"learning_rate": 0.0008947368421052632,
|
| 49 |
+
"loss": 0.9829,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.6,
|
| 54 |
+
"learning_rate": 0.0008796992481203007,
|
| 55 |
+
"loss": 0.9247,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.68,
|
| 60 |
+
"learning_rate": 0.0008646616541353384,
|
| 61 |
+
"loss": 0.9307,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.75,
|
| 66 |
+
"learning_rate": 0.000849624060150376,
|
| 67 |
+
"loss": 0.9512,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.83,
|
| 72 |
+
"learning_rate": 0.0008345864661654136,
|
| 73 |
+
"loss": 0.9202,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.9,
|
| 78 |
+
"learning_rate": 0.0008195488721804511,
|
| 79 |
+
"loss": 0.8894,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.98,
|
| 84 |
+
"learning_rate": 0.0008045112781954887,
|
| 85 |
+
"loss": 0.8794,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 1.05,
|
| 90 |
+
"learning_rate": 0.0007894736842105263,
|
| 91 |
+
"loss": 0.8286,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 1.13,
|
| 96 |
+
"learning_rate": 0.000774436090225564,
|
| 97 |
+
"loss": 0.8827,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.2,
|
| 102 |
+
"learning_rate": 0.0007593984962406016,
|
| 103 |
+
"loss": 0.8604,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 1.28,
|
| 108 |
+
"learning_rate": 0.0007443609022556391,
|
| 109 |
+
"loss": 0.8379,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.35,
|
| 114 |
+
"learning_rate": 0.0007293233082706767,
|
| 115 |
+
"loss": 0.8943,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.43,
|
| 120 |
+
"learning_rate": 0.0007142857142857143,
|
| 121 |
+
"loss": 0.8367,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.5,
|
| 126 |
+
"learning_rate": 0.0006992481203007519,
|
| 127 |
+
"loss": 0.8114,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.58,
|
| 132 |
+
"learning_rate": 0.0006842105263157895,
|
| 133 |
+
"loss": 0.8834,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 1.65,
|
| 138 |
+
"learning_rate": 0.0006691729323308271,
|
| 139 |
+
"loss": 0.8047,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.73,
|
| 144 |
+
"learning_rate": 0.0006541353383458647,
|
| 145 |
+
"loss": 0.8068,
|
| 146 |
+
"step": 230
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.8,
|
| 150 |
+
"learning_rate": 0.0006390977443609023,
|
| 151 |
+
"loss": 0.8402,
|
| 152 |
+
"step": 240
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.88,
|
| 156 |
+
"learning_rate": 0.0006240601503759398,
|
| 157 |
+
"loss": 0.8168,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.95,
|
| 162 |
+
"learning_rate": 0.0006090225563909775,
|
| 163 |
+
"loss": 0.8121,
|
| 164 |
+
"step": 260
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 2.03,
|
| 168 |
+
"learning_rate": 0.0005939849624060151,
|
| 169 |
+
"loss": 0.7975,
|
| 170 |
+
"step": 270
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.11,
|
| 174 |
+
"learning_rate": 0.0005789473684210527,
|
| 175 |
+
"loss": 0.7473,
|
| 176 |
+
"step": 280
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.18,
|
| 180 |
+
"learning_rate": 0.0005639097744360903,
|
| 181 |
+
"loss": 0.7851,
|
| 182 |
+
"step": 290
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 2.26,
|
| 186 |
+
"learning_rate": 0.0005488721804511278,
|
| 187 |
+
"loss": 0.7704,
|
| 188 |
+
"step": 300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.33,
|
| 192 |
+
"learning_rate": 0.0005338345864661654,
|
| 193 |
+
"loss": 0.714,
|
| 194 |
+
"step": 310
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.41,
|
| 198 |
+
"learning_rate": 0.0005187969924812031,
|
| 199 |
+
"loss": 0.6813,
|
| 200 |
+
"step": 320
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.48,
|
| 204 |
+
"learning_rate": 0.0005037593984962407,
|
| 205 |
+
"loss": 0.7964,
|
| 206 |
+
"step": 330
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.56,
|
| 210 |
+
"learning_rate": 0.0004887218045112781,
|
| 211 |
+
"loss": 0.7725,
|
| 212 |
+
"step": 340
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.63,
|
| 216 |
+
"learning_rate": 0.00047368421052631577,
|
| 217 |
+
"loss": 0.684,
|
| 218 |
+
"step": 350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.71,
|
| 222 |
+
"learning_rate": 0.00045864661654135334,
|
| 223 |
+
"loss": 0.7863,
|
| 224 |
+
"step": 360
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 2.78,
|
| 228 |
+
"learning_rate": 0.000443609022556391,
|
| 229 |
+
"loss": 0.7148,
|
| 230 |
+
"step": 370
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.86,
|
| 234 |
+
"learning_rate": 0.00042857142857142855,
|
| 235 |
+
"loss": 0.7665,
|
| 236 |
+
"step": 380
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.93,
|
| 240 |
+
"learning_rate": 0.0004135338345864661,
|
| 241 |
+
"loss": 0.6897,
|
| 242 |
+
"step": 390
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 3.01,
|
| 246 |
+
"learning_rate": 0.00039849624060150376,
|
| 247 |
+
"loss": 0.7275,
|
| 248 |
+
"step": 400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 3.08,
|
| 252 |
+
"learning_rate": 0.00038345864661654133,
|
| 253 |
+
"loss": 0.7285,
|
| 254 |
+
"step": 410
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 3.16,
|
| 258 |
+
"learning_rate": 0.00036842105263157896,
|
| 259 |
+
"loss": 0.6366,
|
| 260 |
+
"step": 420
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 3.23,
|
| 264 |
+
"learning_rate": 0.00035338345864661654,
|
| 265 |
+
"loss": 0.6528,
|
| 266 |
+
"step": 430
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 3.31,
|
| 270 |
+
"learning_rate": 0.0003383458646616541,
|
| 271 |
+
"loss": 0.5579,
|
| 272 |
+
"step": 440
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 3.38,
|
| 276 |
+
"learning_rate": 0.00032330827067669174,
|
| 277 |
+
"loss": 0.6154,
|
| 278 |
+
"step": 450
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 3.46,
|
| 282 |
+
"learning_rate": 0.0003082706766917293,
|
| 283 |
+
"loss": 0.6,
|
| 284 |
+
"step": 460
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 3.53,
|
| 288 |
+
"learning_rate": 0.0002932330827067669,
|
| 289 |
+
"loss": 0.632,
|
| 290 |
+
"step": 470
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 3.61,
|
| 294 |
+
"learning_rate": 0.0002781954887218045,
|
| 295 |
+
"loss": 0.6108,
|
| 296 |
+
"step": 480
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 3.68,
|
| 300 |
+
"learning_rate": 0.0002631578947368421,
|
| 301 |
+
"loss": 0.6173,
|
| 302 |
+
"step": 490
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 3.76,
|
| 306 |
+
"learning_rate": 0.0002481203007518797,
|
| 307 |
+
"loss": 0.6389,
|
| 308 |
+
"step": 500
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 3.83,
|
| 312 |
+
"learning_rate": 0.00023308270676691728,
|
| 313 |
+
"loss": 0.5448,
|
| 314 |
+
"step": 510
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 3.91,
|
| 318 |
+
"learning_rate": 0.00021804511278195488,
|
| 319 |
+
"loss": 0.5878,
|
| 320 |
+
"step": 520
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 3.98,
|
| 324 |
+
"learning_rate": 0.00020300751879699248,
|
| 325 |
+
"loss": 0.5221,
|
| 326 |
+
"step": 530
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 4.06,
|
| 330 |
+
"learning_rate": 0.00018796992481203006,
|
| 331 |
+
"loss": 0.4697,
|
| 332 |
+
"step": 540
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 4.14,
|
| 336 |
+
"learning_rate": 0.00017293233082706766,
|
| 337 |
+
"loss": 0.466,
|
| 338 |
+
"step": 550
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 4.21,
|
| 342 |
+
"learning_rate": 0.00015789473684210527,
|
| 343 |
+
"loss": 0.5204,
|
| 344 |
+
"step": 560
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.29,
|
| 348 |
+
"learning_rate": 0.00014285714285714284,
|
| 349 |
+
"loss": 0.4632,
|
| 350 |
+
"step": 570
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 4.36,
|
| 354 |
+
"learning_rate": 0.00012781954887218044,
|
| 355 |
+
"loss": 0.4604,
|
| 356 |
+
"step": 580
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 4.44,
|
| 360 |
+
"learning_rate": 0.00011278195488721805,
|
| 361 |
+
"loss": 0.4924,
|
| 362 |
+
"step": 590
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 4.51,
|
| 366 |
+
"learning_rate": 9.774436090225564e-05,
|
| 367 |
+
"loss": 0.4671,
|
| 368 |
+
"step": 600
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 4.59,
|
| 372 |
+
"learning_rate": 8.270676691729323e-05,
|
| 373 |
+
"loss": 0.5123,
|
| 374 |
+
"step": 610
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 4.66,
|
| 378 |
+
"learning_rate": 6.766917293233083e-05,
|
| 379 |
+
"loss": 0.5051,
|
| 380 |
+
"step": 620
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 4.74,
|
| 384 |
+
"learning_rate": 5.263157894736842e-05,
|
| 385 |
+
"loss": 0.4642,
|
| 386 |
+
"step": 630
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 4.81,
|
| 390 |
+
"learning_rate": 3.759398496240601e-05,
|
| 391 |
+
"loss": 0.4851,
|
| 392 |
+
"step": 640
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 4.89,
|
| 396 |
+
"learning_rate": 2.255639097744361e-05,
|
| 397 |
+
"loss": 0.4467,
|
| 398 |
+
"step": 650
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 4.96,
|
| 402 |
+
"learning_rate": 7.518796992481203e-06,
|
| 403 |
+
"loss": 0.4283,
|
| 404 |
+
"step": 660
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 5.0,
|
| 408 |
+
"step": 665,
|
| 409 |
+
"total_flos": 24976042475520.0,
|
| 410 |
+
"train_loss": 0.7294828687395368,
|
| 411 |
+
"train_runtime": 22.1174,
|
| 412 |
+
"train_samples_per_second": 956.715,
|
| 413 |
+
"train_steps_per_second": 30.067
|
| 414 |
+
}
|
| 415 |
+
],
|
| 416 |
+
"max_steps": 665,
|
| 417 |
+
"num_train_epochs": 5,
|
| 418 |
+
"total_flos": 24976042475520.0,
|
| 419 |
+
"trial_name": null,
|
| 420 |
+
"trial_params": null
|
| 421 |
+
}
|