Commit ·
f707cb1
1
Parent(s): ba98d3e
Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +412 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,412 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.18451988697052002,
|
| 3 |
+
"best_model_checkpoint": "./models/results_one_liners_693/checkpoint-325",
|
| 4 |
+
"epoch": 0.9374436632413917,
|
| 5 |
+
"global_step": 325,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.02,
|
| 12 |
+
"learning_rate": 4.000000000000001e-06,
|
| 13 |
+
"loss": 0.6947,
|
| 14 |
+
"step": 8
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"learning_rate": 8.000000000000001e-06,
|
| 19 |
+
"loss": 0.6932,
|
| 20 |
+
"step": 16
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.07,
|
| 24 |
+
"learning_rate": 1.2e-05,
|
| 25 |
+
"loss": 0.6919,
|
| 26 |
+
"step": 24
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.07,
|
| 30 |
+
"eval_accuracy": 0.6413249211356467,
|
| 31 |
+
"eval_f1": 0.7321554770318023,
|
| 32 |
+
"eval_loss": 0.6814908981323242,
|
| 33 |
+
"eval_precision": 0.5779100037188546,
|
| 34 |
+
"eval_recall": 0.9987146529562982,
|
| 35 |
+
"eval_runtime": 21.2769,
|
| 36 |
+
"eval_samples_per_second": 148.988,
|
| 37 |
+
"eval_steps_per_second": 18.659,
|
| 38 |
+
"step": 25
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.09,
|
| 42 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 43 |
+
"loss": 0.673,
|
| 44 |
+
"step": 32
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.12,
|
| 48 |
+
"learning_rate": 2e-05,
|
| 49 |
+
"loss": 0.6191,
|
| 50 |
+
"step": 40
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.14,
|
| 54 |
+
"learning_rate": 2.4e-05,
|
| 55 |
+
"loss": 0.4553,
|
| 56 |
+
"step": 48
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.14,
|
| 60 |
+
"eval_accuracy": 0.8596214511041009,
|
| 61 |
+
"eval_f1": 0.8499156829679595,
|
| 62 |
+
"eval_loss": 0.34521493315696716,
|
| 63 |
+
"eval_precision": 0.8942512420156139,
|
| 64 |
+
"eval_recall": 0.8097686375321337,
|
| 65 |
+
"eval_runtime": 20.9603,
|
| 66 |
+
"eval_samples_per_second": 151.238,
|
| 67 |
+
"eval_steps_per_second": 18.941,
|
| 68 |
+
"step": 50
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.16,
|
| 72 |
+
"learning_rate": 2.8000000000000003e-05,
|
| 73 |
+
"loss": 0.344,
|
| 74 |
+
"step": 56
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.18,
|
| 78 |
+
"learning_rate": 3.2000000000000005e-05,
|
| 79 |
+
"loss": 0.3474,
|
| 80 |
+
"step": 64
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.21,
|
| 84 |
+
"learning_rate": 3.6e-05,
|
| 85 |
+
"loss": 0.363,
|
| 86 |
+
"step": 72
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.22,
|
| 90 |
+
"eval_accuracy": 0.8933753943217666,
|
| 91 |
+
"eval_f1": 0.8901884340480831,
|
| 92 |
+
"eval_loss": 0.25991594791412354,
|
| 93 |
+
"eval_precision": 0.900131406044678,
|
| 94 |
+
"eval_recall": 0.8804627249357326,
|
| 95 |
+
"eval_runtime": 20.8982,
|
| 96 |
+
"eval_samples_per_second": 151.688,
|
| 97 |
+
"eval_steps_per_second": 18.997,
|
| 98 |
+
"step": 75
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.23,
|
| 102 |
+
"learning_rate": 3.9500000000000005e-05,
|
| 103 |
+
"loss": 0.3698,
|
| 104 |
+
"step": 80
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.25,
|
| 108 |
+
"learning_rate": 4.35e-05,
|
| 109 |
+
"loss": 0.2855,
|
| 110 |
+
"step": 88
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.28,
|
| 114 |
+
"learning_rate": 4.75e-05,
|
| 115 |
+
"loss": 0.3564,
|
| 116 |
+
"step": 96
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.29,
|
| 120 |
+
"eval_accuracy": 0.8902208201892744,
|
| 121 |
+
"eval_f1": 0.8932515337423313,
|
| 122 |
+
"eval_loss": 0.2616112232208252,
|
| 123 |
+
"eval_precision": 0.8544600938967136,
|
| 124 |
+
"eval_recall": 0.9357326478149101,
|
| 125 |
+
"eval_runtime": 20.9191,
|
| 126 |
+
"eval_samples_per_second": 151.536,
|
| 127 |
+
"eval_steps_per_second": 18.978,
|
| 128 |
+
"step": 100
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.3,
|
| 132 |
+
"learning_rate": 4.9390243902439024e-05,
|
| 133 |
+
"loss": 0.271,
|
| 134 |
+
"step": 104
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.32,
|
| 138 |
+
"learning_rate": 4.776422764227643e-05,
|
| 139 |
+
"loss": 0.3519,
|
| 140 |
+
"step": 112
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.35,
|
| 144 |
+
"learning_rate": 4.613821138211382e-05,
|
| 145 |
+
"loss": 0.3108,
|
| 146 |
+
"step": 120
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.36,
|
| 150 |
+
"eval_accuracy": 0.8943217665615142,
|
| 151 |
+
"eval_f1": 0.8949513954217623,
|
| 152 |
+
"eval_loss": 0.2482648640871048,
|
| 153 |
+
"eval_precision": 0.8738518064911206,
|
| 154 |
+
"eval_recall": 0.9170951156812339,
|
| 155 |
+
"eval_runtime": 20.9132,
|
| 156 |
+
"eval_samples_per_second": 151.579,
|
| 157 |
+
"eval_steps_per_second": 18.983,
|
| 158 |
+
"step": 125
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.37,
|
| 162 |
+
"learning_rate": 4.451219512195122e-05,
|
| 163 |
+
"loss": 0.2794,
|
| 164 |
+
"step": 128
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.39,
|
| 168 |
+
"learning_rate": 4.2886178861788616e-05,
|
| 169 |
+
"loss": 0.3446,
|
| 170 |
+
"step": 136
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.42,
|
| 174 |
+
"learning_rate": 4.146341463414634e-05,
|
| 175 |
+
"loss": 0.2649,
|
| 176 |
+
"step": 144
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.43,
|
| 180 |
+
"eval_accuracy": 0.9123028391167193,
|
| 181 |
+
"eval_f1": 0.9100905562742561,
|
| 182 |
+
"eval_loss": 0.22195443511009216,
|
| 183 |
+
"eval_precision": 0.916015625,
|
| 184 |
+
"eval_recall": 0.9042416452442159,
|
| 185 |
+
"eval_runtime": 20.8531,
|
| 186 |
+
"eval_samples_per_second": 152.016,
|
| 187 |
+
"eval_steps_per_second": 19.038,
|
| 188 |
+
"step": 150
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.44,
|
| 192 |
+
"learning_rate": 3.983739837398374e-05,
|
| 193 |
+
"loss": 0.2772,
|
| 194 |
+
"step": 152
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.46,
|
| 198 |
+
"learning_rate": 3.8211382113821145e-05,
|
| 199 |
+
"loss": 0.2738,
|
| 200 |
+
"step": 160
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.48,
|
| 204 |
+
"learning_rate": 3.6585365853658535e-05,
|
| 205 |
+
"loss": 0.3146,
|
| 206 |
+
"step": 168
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.5,
|
| 210 |
+
"eval_accuracy": 0.9082018927444795,
|
| 211 |
+
"eval_f1": 0.9041817583141256,
|
| 212 |
+
"eval_loss": 0.22258049249649048,
|
| 213 |
+
"eval_precision": 0.9270762997974341,
|
| 214 |
+
"eval_recall": 0.8823907455012854,
|
| 215 |
+
"eval_runtime": 20.9525,
|
| 216 |
+
"eval_samples_per_second": 151.295,
|
| 217 |
+
"eval_steps_per_second": 18.948,
|
| 218 |
+
"step": 175
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.51,
|
| 222 |
+
"learning_rate": 3.495934959349594e-05,
|
| 223 |
+
"loss": 0.2606,
|
| 224 |
+
"step": 176
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.53,
|
| 228 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 229 |
+
"loss": 0.1921,
|
| 230 |
+
"step": 184
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.55,
|
| 234 |
+
"learning_rate": 3.170731707317073e-05,
|
| 235 |
+
"loss": 0.1973,
|
| 236 |
+
"step": 192
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.58,
|
| 240 |
+
"learning_rate": 3.0081300813008135e-05,
|
| 241 |
+
"loss": 0.2459,
|
| 242 |
+
"step": 200
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.58,
|
| 246 |
+
"eval_accuracy": 0.8917981072555206,
|
| 247 |
+
"eval_f1": 0.8987304399173309,
|
| 248 |
+
"eval_loss": 0.27942967414855957,
|
| 249 |
+
"eval_precision": 0.8312397596941562,
|
| 250 |
+
"eval_recall": 0.9781491002570694,
|
| 251 |
+
"eval_runtime": 21.1312,
|
| 252 |
+
"eval_samples_per_second": 150.015,
|
| 253 |
+
"eval_steps_per_second": 18.787,
|
| 254 |
+
"step": 200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.6,
|
| 258 |
+
"learning_rate": 2.8455284552845528e-05,
|
| 259 |
+
"loss": 0.2524,
|
| 260 |
+
"step": 208
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.62,
|
| 264 |
+
"learning_rate": 2.682926829268293e-05,
|
| 265 |
+
"loss": 0.2249,
|
| 266 |
+
"step": 216
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.65,
|
| 270 |
+
"learning_rate": 2.5203252032520324e-05,
|
| 271 |
+
"loss": 0.286,
|
| 272 |
+
"step": 224
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.65,
|
| 276 |
+
"eval_accuracy": 0.9261829652996846,
|
| 277 |
+
"eval_f1": 0.9265536723163842,
|
| 278 |
+
"eval_loss": 0.20206698775291443,
|
| 279 |
+
"eval_precision": 0.905521472392638,
|
| 280 |
+
"eval_recall": 0.9485861182519281,
|
| 281 |
+
"eval_runtime": 21.0759,
|
| 282 |
+
"eval_samples_per_second": 150.409,
|
| 283 |
+
"eval_steps_per_second": 18.837,
|
| 284 |
+
"step": 225
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.67,
|
| 288 |
+
"learning_rate": 2.3577235772357724e-05,
|
| 289 |
+
"loss": 0.2469,
|
| 290 |
+
"step": 232
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.69,
|
| 294 |
+
"learning_rate": 2.1951219512195124e-05,
|
| 295 |
+
"loss": 0.2506,
|
| 296 |
+
"step": 240
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.72,
|
| 300 |
+
"learning_rate": 2.032520325203252e-05,
|
| 301 |
+
"loss": 0.1756,
|
| 302 |
+
"step": 248
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.72,
|
| 306 |
+
"eval_accuracy": 0.9233438485804416,
|
| 307 |
+
"eval_f1": 0.9235127478753541,
|
| 308 |
+
"eval_loss": 0.20944611728191376,
|
| 309 |
+
"eval_precision": 0.904996915484269,
|
| 310 |
+
"eval_recall": 0.9428020565552699,
|
| 311 |
+
"eval_runtime": 21.6836,
|
| 312 |
+
"eval_samples_per_second": 146.193,
|
| 313 |
+
"eval_steps_per_second": 18.309,
|
| 314 |
+
"step": 250
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.74,
|
| 318 |
+
"learning_rate": 1.869918699186992e-05,
|
| 319 |
+
"loss": 0.2227,
|
| 320 |
+
"step": 256
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.76,
|
| 324 |
+
"learning_rate": 1.707317073170732e-05,
|
| 325 |
+
"loss": 0.2191,
|
| 326 |
+
"step": 264
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.78,
|
| 330 |
+
"learning_rate": 1.5447154471544717e-05,
|
| 331 |
+
"loss": 0.1611,
|
| 332 |
+
"step": 272
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.79,
|
| 336 |
+
"eval_accuracy": 0.9280757097791799,
|
| 337 |
+
"eval_f1": 0.9263089851325145,
|
| 338 |
+
"eval_loss": 0.19648678600788116,
|
| 339 |
+
"eval_precision": 0.9317295188556567,
|
| 340 |
+
"eval_recall": 0.9209511568123393,
|
| 341 |
+
"eval_runtime": 21.0389,
|
| 342 |
+
"eval_samples_per_second": 150.673,
|
| 343 |
+
"eval_steps_per_second": 18.87,
|
| 344 |
+
"step": 275
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.81,
|
| 348 |
+
"learning_rate": 1.3821138211382115e-05,
|
| 349 |
+
"loss": 0.1854,
|
| 350 |
+
"step": 280
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.83,
|
| 354 |
+
"learning_rate": 1.2195121951219513e-05,
|
| 355 |
+
"loss": 0.1772,
|
| 356 |
+
"step": 288
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.85,
|
| 360 |
+
"learning_rate": 1.0569105691056911e-05,
|
| 361 |
+
"loss": 0.2562,
|
| 362 |
+
"step": 296
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.87,
|
| 366 |
+
"eval_accuracy": 0.9236593059936908,
|
| 367 |
+
"eval_f1": 0.9252625077208154,
|
| 368 |
+
"eval_loss": 0.20132240653038025,
|
| 369 |
+
"eval_precision": 0.8906064209274673,
|
| 370 |
+
"eval_recall": 0.9627249357326478,
|
| 371 |
+
"eval_runtime": 21.0921,
|
| 372 |
+
"eval_samples_per_second": 150.293,
|
| 373 |
+
"eval_steps_per_second": 18.822,
|
| 374 |
+
"step": 300
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.88,
|
| 378 |
+
"learning_rate": 8.94308943089431e-06,
|
| 379 |
+
"loss": 0.2303,
|
| 380 |
+
"step": 304
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.9,
|
| 384 |
+
"learning_rate": 7.317073170731707e-06,
|
| 385 |
+
"loss": 0.2242,
|
| 386 |
+
"step": 312
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.92,
|
| 390 |
+
"learning_rate": 5.6910569105691056e-06,
|
| 391 |
+
"loss": 0.1955,
|
| 392 |
+
"step": 320
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.94,
|
| 396 |
+
"eval_accuracy": 0.931230283911672,
|
| 397 |
+
"eval_f1": 0.9316185696361355,
|
| 398 |
+
"eval_loss": 0.18451988697052002,
|
| 399 |
+
"eval_precision": 0.9099264705882353,
|
| 400 |
+
"eval_recall": 0.9543701799485861,
|
| 401 |
+
"eval_runtime": 20.9834,
|
| 402 |
+
"eval_samples_per_second": 151.072,
|
| 403 |
+
"eval_steps_per_second": 18.92,
|
| 404 |
+
"step": 325
|
| 405 |
+
}
|
| 406 |
+
],
|
| 407 |
+
"max_steps": 346,
|
| 408 |
+
"num_train_epochs": 1,
|
| 409 |
+
"total_flos": 619955424192000.0,
|
| 410 |
+
"trial_name": null,
|
| 411 |
+
"trial_params": null
|
| 412 |
+
}
|