./output/gpt2.CEBaB_confounding.price_food_ambiance_negative.absa.5-class.seed_42
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": 0.8294864029498022,
|
| 4 |
+
"eval_Weighted-Macro-F1": 0.8313025665825312,
|
| 5 |
+
"eval_accuracy": 0.8310893512851897,
|
| 6 |
+
"eval_loss": 0.4725877642631531,
|
| 7 |
+
"eval_runtime": 13.675,
|
| 8 |
+
"eval_samples": 4902,
|
| 9 |
+
"eval_samples_per_second": 358.463,
|
| 10 |
+
"eval_steps_per_second": 11.261,
|
| 11 |
+
"train_loss": 0.6022974249088403,
|
| 12 |
+
"train_runtime": 176.6939,
|
| 13 |
+
"train_samples": 4207,
|
| 14 |
+
"train_samples_per_second": 119.048,
|
| 15 |
+
"train_steps_per_second": 3.735
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_Macro-F1": 0.8294864029498022,
|
| 4 |
+
"eval_Weighted-Macro-F1": 0.8313025665825312,
|
| 5 |
+
"eval_accuracy": 0.8310893512851897,
|
| 6 |
+
"eval_loss": 0.4725877642631531,
|
| 7 |
+
"eval_runtime": 13.675,
|
| 8 |
+
"eval_samples": 4902,
|
| 9 |
+
"eval_samples_per_second": 358.463,
|
| 10 |
+
"eval_steps_per_second": 11.261
|
| 11 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"train_loss": 0.6022974249088403,
|
| 4 |
+
"train_runtime": 176.6939,
|
| 5 |
+
"train_samples": 4207,
|
| 6 |
+
"train_samples_per_second": 119.048,
|
| 7 |
+
"train_steps_per_second": 3.735
|
| 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": 660,
|
| 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": 1.96969696969697e-05,
|
| 13 |
+
"loss": 1.1658,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.15,
|
| 18 |
+
"learning_rate": 1.9393939393939395e-05,
|
| 19 |
+
"loss": 1.1076,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.23,
|
| 24 |
+
"learning_rate": 1.9090909090909094e-05,
|
| 25 |
+
"loss": 1.0267,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.3,
|
| 30 |
+
"learning_rate": 1.8787878787878792e-05,
|
| 31 |
+
"loss": 0.9707,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.38,
|
| 36 |
+
"learning_rate": 1.8484848484848487e-05,
|
| 37 |
+
"loss": 1.0231,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.45,
|
| 42 |
+
"learning_rate": 1.8181818181818182e-05,
|
| 43 |
+
"loss": 0.9652,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.53,
|
| 48 |
+
"learning_rate": 1.787878787878788e-05,
|
| 49 |
+
"loss": 0.9454,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.61,
|
| 54 |
+
"learning_rate": 1.7575757575757576e-05,
|
| 55 |
+
"loss": 0.9635,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.68,
|
| 60 |
+
"learning_rate": 1.7272727272727274e-05,
|
| 61 |
+
"loss": 0.9535,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.76,
|
| 66 |
+
"learning_rate": 1.6969696969696972e-05,
|
| 67 |
+
"loss": 0.9017,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.83,
|
| 72 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 73 |
+
"loss": 0.8334,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.91,
|
| 78 |
+
"learning_rate": 1.6363636363636366e-05,
|
| 79 |
+
"loss": 0.8924,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.98,
|
| 84 |
+
"learning_rate": 1.606060606060606e-05,
|
| 85 |
+
"loss": 0.8255,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 1.06,
|
| 90 |
+
"learning_rate": 1.575757575757576e-05,
|
| 91 |
+
"loss": 0.7972,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 1.14,
|
| 96 |
+
"learning_rate": 1.5454545454545454e-05,
|
| 97 |
+
"loss": 0.8028,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.21,
|
| 102 |
+
"learning_rate": 1.5151515151515153e-05,
|
| 103 |
+
"loss": 0.7566,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 1.29,
|
| 108 |
+
"learning_rate": 1.484848484848485e-05,
|
| 109 |
+
"loss": 0.7345,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.36,
|
| 114 |
+
"learning_rate": 1.4545454545454546e-05,
|
| 115 |
+
"loss": 0.7123,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.44,
|
| 120 |
+
"learning_rate": 1.4242424242424245e-05,
|
| 121 |
+
"loss": 0.7344,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.52,
|
| 126 |
+
"learning_rate": 1.3939393939393942e-05,
|
| 127 |
+
"loss": 0.6757,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.59,
|
| 132 |
+
"learning_rate": 1.3636363636363637e-05,
|
| 133 |
+
"loss": 0.665,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 1.67,
|
| 138 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 139 |
+
"loss": 0.7071,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.74,
|
| 144 |
+
"learning_rate": 1.3030303030303032e-05,
|
| 145 |
+
"loss": 0.6266,
|
| 146 |
+
"step": 230
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.82,
|
| 150 |
+
"learning_rate": 1.2727272727272728e-05,
|
| 151 |
+
"loss": 0.5948,
|
| 152 |
+
"step": 240
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.89,
|
| 156 |
+
"learning_rate": 1.2424242424242425e-05,
|
| 157 |
+
"loss": 0.6127,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.97,
|
| 162 |
+
"learning_rate": 1.2121212121212122e-05,
|
| 163 |
+
"loss": 0.5923,
|
| 164 |
+
"step": 260
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 2.05,
|
| 168 |
+
"learning_rate": 1.181818181818182e-05,
|
| 169 |
+
"loss": 0.5744,
|
| 170 |
+
"step": 270
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.12,
|
| 174 |
+
"learning_rate": 1.1515151515151517e-05,
|
| 175 |
+
"loss": 0.6321,
|
| 176 |
+
"step": 280
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.2,
|
| 180 |
+
"learning_rate": 1.1212121212121212e-05,
|
| 181 |
+
"loss": 0.4866,
|
| 182 |
+
"step": 290
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 2.27,
|
| 186 |
+
"learning_rate": 1.0909090909090909e-05,
|
| 187 |
+
"loss": 0.4829,
|
| 188 |
+
"step": 300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.35,
|
| 192 |
+
"learning_rate": 1.0606060606060606e-05,
|
| 193 |
+
"loss": 0.5416,
|
| 194 |
+
"step": 310
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.42,
|
| 198 |
+
"learning_rate": 1.0303030303030304e-05,
|
| 199 |
+
"loss": 0.5801,
|
| 200 |
+
"step": 320
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.5,
|
| 204 |
+
"learning_rate": 1e-05,
|
| 205 |
+
"loss": 0.536,
|
| 206 |
+
"step": 330
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.58,
|
| 210 |
+
"learning_rate": 9.696969696969698e-06,
|
| 211 |
+
"loss": 0.4988,
|
| 212 |
+
"step": 340
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.65,
|
| 216 |
+
"learning_rate": 9.393939393939396e-06,
|
| 217 |
+
"loss": 0.5227,
|
| 218 |
+
"step": 350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.73,
|
| 222 |
+
"learning_rate": 9.090909090909091e-06,
|
| 223 |
+
"loss": 0.5278,
|
| 224 |
+
"step": 360
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 2.8,
|
| 228 |
+
"learning_rate": 8.787878787878788e-06,
|
| 229 |
+
"loss": 0.5426,
|
| 230 |
+
"step": 370
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.88,
|
| 234 |
+
"learning_rate": 8.484848484848486e-06,
|
| 235 |
+
"loss": 0.5464,
|
| 236 |
+
"step": 380
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.95,
|
| 240 |
+
"learning_rate": 8.181818181818183e-06,
|
| 241 |
+
"loss": 0.415,
|
| 242 |
+
"step": 390
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 3.03,
|
| 246 |
+
"learning_rate": 7.87878787878788e-06,
|
| 247 |
+
"loss": 0.4644,
|
| 248 |
+
"step": 400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 3.11,
|
| 252 |
+
"learning_rate": 7.5757575757575764e-06,
|
| 253 |
+
"loss": 0.4674,
|
| 254 |
+
"step": 410
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 3.18,
|
| 258 |
+
"learning_rate": 7.272727272727273e-06,
|
| 259 |
+
"loss": 0.441,
|
| 260 |
+
"step": 420
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 3.26,
|
| 264 |
+
"learning_rate": 6.969696969696971e-06,
|
| 265 |
+
"loss": 0.5166,
|
| 266 |
+
"step": 430
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 3.33,
|
| 270 |
+
"learning_rate": 6.666666666666667e-06,
|
| 271 |
+
"loss": 0.4607,
|
| 272 |
+
"step": 440
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 3.41,
|
| 276 |
+
"learning_rate": 6.363636363636364e-06,
|
| 277 |
+
"loss": 0.428,
|
| 278 |
+
"step": 450
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 3.48,
|
| 282 |
+
"learning_rate": 6.060606060606061e-06,
|
| 283 |
+
"loss": 0.4172,
|
| 284 |
+
"step": 460
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 3.56,
|
| 288 |
+
"learning_rate": 5.7575757575757586e-06,
|
| 289 |
+
"loss": 0.4494,
|
| 290 |
+
"step": 470
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 3.64,
|
| 294 |
+
"learning_rate": 5.4545454545454545e-06,
|
| 295 |
+
"loss": 0.4084,
|
| 296 |
+
"step": 480
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 3.71,
|
| 300 |
+
"learning_rate": 5.151515151515152e-06,
|
| 301 |
+
"loss": 0.4428,
|
| 302 |
+
"step": 490
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 3.79,
|
| 306 |
+
"learning_rate": 4.848484848484849e-06,
|
| 307 |
+
"loss": 0.4007,
|
| 308 |
+
"step": 500
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 3.86,
|
| 312 |
+
"learning_rate": 4.5454545454545455e-06,
|
| 313 |
+
"loss": 0.4297,
|
| 314 |
+
"step": 510
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 3.94,
|
| 318 |
+
"learning_rate": 4.242424242424243e-06,
|
| 319 |
+
"loss": 0.4108,
|
| 320 |
+
"step": 520
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 4.02,
|
| 324 |
+
"learning_rate": 3.93939393939394e-06,
|
| 325 |
+
"loss": 0.4433,
|
| 326 |
+
"step": 530
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 4.09,
|
| 330 |
+
"learning_rate": 3.6363636363636366e-06,
|
| 331 |
+
"loss": 0.4151,
|
| 332 |
+
"step": 540
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 4.17,
|
| 336 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 337 |
+
"loss": 0.3749,
|
| 338 |
+
"step": 550
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 4.24,
|
| 342 |
+
"learning_rate": 3.0303030303030305e-06,
|
| 343 |
+
"loss": 0.3343,
|
| 344 |
+
"step": 560
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.32,
|
| 348 |
+
"learning_rate": 2.7272727272727272e-06,
|
| 349 |
+
"loss": 0.3746,
|
| 350 |
+
"step": 570
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 4.39,
|
| 354 |
+
"learning_rate": 2.4242424242424244e-06,
|
| 355 |
+
"loss": 0.4396,
|
| 356 |
+
"step": 580
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 4.47,
|
| 360 |
+
"learning_rate": 2.1212121212121216e-06,
|
| 361 |
+
"loss": 0.3793,
|
| 362 |
+
"step": 590
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 4.55,
|
| 366 |
+
"learning_rate": 1.8181818181818183e-06,
|
| 367 |
+
"loss": 0.4193,
|
| 368 |
+
"step": 600
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 4.62,
|
| 372 |
+
"learning_rate": 1.5151515151515152e-06,
|
| 373 |
+
"loss": 0.4185,
|
| 374 |
+
"step": 610
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 4.7,
|
| 378 |
+
"learning_rate": 1.2121212121212122e-06,
|
| 379 |
+
"loss": 0.3987,
|
| 380 |
+
"step": 620
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 4.77,
|
| 384 |
+
"learning_rate": 9.090909090909091e-07,
|
| 385 |
+
"loss": 0.3946,
|
| 386 |
+
"step": 630
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 4.85,
|
| 390 |
+
"learning_rate": 6.060606060606061e-07,
|
| 391 |
+
"loss": 0.3512,
|
| 392 |
+
"step": 640
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 4.92,
|
| 396 |
+
"learning_rate": 3.0303030303030305e-07,
|
| 397 |
+
"loss": 0.3771,
|
| 398 |
+
"step": 650
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 5.0,
|
| 402 |
+
"learning_rate": 0.0,
|
| 403 |
+
"loss": 0.4206,
|
| 404 |
+
"step": 660
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 5.0,
|
| 408 |
+
"step": 660,
|
| 409 |
+
"total_flos": 1383647685477120.0,
|
| 410 |
+
"train_loss": 0.6022974249088403,
|
| 411 |
+
"train_runtime": 176.6939,
|
| 412 |
+
"train_samples_per_second": 119.048,
|
| 413 |
+
"train_steps_per_second": 3.735
|
| 414 |
+
}
|
| 415 |
+
],
|
| 416 |
+
"max_steps": 660,
|
| 417 |
+
"num_train_epochs": 5,
|
| 418 |
+
"total_flos": 1383647685477120.0,
|
| 419 |
+
"trial_name": null,
|
| 420 |
+
"trial_params": null
|
| 421 |
+
}
|