Training in progress, epoch 1
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 2118227382190080.0,
|
| 4 |
+
"train_loss": 0.374498660454969,
|
| 5 |
+
"train_runtime": 306.4673,
|
| 6 |
+
"train_samples_per_second": 52.156,
|
| 7 |
+
"train_steps_per_second": 1.635
|
| 8 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 267926253
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:748b0b72035176a7410c96119112e2170d5c4ea7ee975b13d3215acf6953fc77
|
| 3 |
size 267926253
|
runs/Mar10_12-04-59_96c678b480cf/1678452496.7788405/events.out.tfevents.1678452496.96c678b480cf.1488.5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:907cf02f0b10ab03f9a4e52b52691b1e4535d907a3182db14112621286107e23
|
| 3 |
+
size 5767
|
runs/Mar10_12-04-59_96c678b480cf/events.out.tfevents.1678452496.96c678b480cf.1488.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8ad2ba5894ef7d8b4c0572be962af283481da7ca5511edff23c7938586d7ed10
|
| 3 |
+
size 7880
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 2118227382190080.0,
|
| 4 |
+
"train_loss": 0.374498660454969,
|
| 5 |
+
"train_runtime": 306.4673,
|
| 6 |
+
"train_samples_per_second": 52.156,
|
| 7 |
+
"train_steps_per_second": 1.635
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,361 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9849327674451294,
|
| 3 |
+
"best_model_checkpoint": "distilbert-base-uncased_finetuned_text_2_disease/checkpoint-167",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"global_step": 501,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.06,
|
| 12 |
+
"learning_rate": 0.0,
|
| 13 |
+
"loss": 0.3927,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.12,
|
| 18 |
+
"learning_rate": 0.0,
|
| 19 |
+
"loss": 0.3871,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.18,
|
| 24 |
+
"learning_rate": 0.0,
|
| 25 |
+
"loss": 0.3545,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.24,
|
| 30 |
+
"learning_rate": 0.0,
|
| 31 |
+
"loss": 0.389,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.3,
|
| 36 |
+
"learning_rate": 0.0,
|
| 37 |
+
"loss": 0.3796,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.36,
|
| 42 |
+
"learning_rate": 0.0,
|
| 43 |
+
"loss": 0.3484,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.42,
|
| 48 |
+
"learning_rate": 0.0,
|
| 49 |
+
"loss": 0.3691,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.48,
|
| 54 |
+
"learning_rate": 0.0,
|
| 55 |
+
"loss": 0.3773,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.54,
|
| 60 |
+
"learning_rate": 0.0,
|
| 61 |
+
"loss": 0.3739,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.6,
|
| 66 |
+
"learning_rate": 0.0,
|
| 67 |
+
"loss": 0.3687,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.66,
|
| 72 |
+
"learning_rate": 0.0,
|
| 73 |
+
"loss": 0.3946,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.72,
|
| 78 |
+
"learning_rate": 0.0,
|
| 79 |
+
"loss": 0.3614,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.78,
|
| 84 |
+
"learning_rate": 0.0,
|
| 85 |
+
"loss": 0.3963,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.84,
|
| 90 |
+
"learning_rate": 0.0,
|
| 91 |
+
"loss": 0.3555,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.9,
|
| 96 |
+
"learning_rate": 0.0,
|
| 97 |
+
"loss": 0.3627,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.96,
|
| 102 |
+
"learning_rate": 0.0,
|
| 103 |
+
"loss": 0.3641,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 1.0,
|
| 108 |
+
"eval_accuracy": 0.984984984984985,
|
| 109 |
+
"eval_f1": 0.9849327674451294,
|
| 110 |
+
"eval_loss": 0.2975810170173645,
|
| 111 |
+
"eval_precision": 0.9865442976554087,
|
| 112 |
+
"eval_recall": 0.984984984984985,
|
| 113 |
+
"eval_runtime": 3.8994,
|
| 114 |
+
"eval_samples_per_second": 170.795,
|
| 115 |
+
"eval_steps_per_second": 5.385,
|
| 116 |
+
"step": 167
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.02,
|
| 120 |
+
"learning_rate": 0.0,
|
| 121 |
+
"loss": 0.3898,
|
| 122 |
+
"step": 170
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.08,
|
| 126 |
+
"learning_rate": 0.0,
|
| 127 |
+
"loss": 0.3551,
|
| 128 |
+
"step": 180
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.14,
|
| 132 |
+
"learning_rate": 0.0,
|
| 133 |
+
"loss": 0.3864,
|
| 134 |
+
"step": 190
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 1.2,
|
| 138 |
+
"learning_rate": 0.0,
|
| 139 |
+
"loss": 0.3539,
|
| 140 |
+
"step": 200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.26,
|
| 144 |
+
"learning_rate": 0.0,
|
| 145 |
+
"loss": 0.3792,
|
| 146 |
+
"step": 210
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.32,
|
| 150 |
+
"learning_rate": 0.0,
|
| 151 |
+
"loss": 0.3929,
|
| 152 |
+
"step": 220
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.38,
|
| 156 |
+
"learning_rate": 0.0,
|
| 157 |
+
"loss": 0.3688,
|
| 158 |
+
"step": 230
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.44,
|
| 162 |
+
"learning_rate": 0.0,
|
| 163 |
+
"loss": 0.4244,
|
| 164 |
+
"step": 240
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 1.5,
|
| 168 |
+
"learning_rate": 0.0,
|
| 169 |
+
"loss": 0.3677,
|
| 170 |
+
"step": 250
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.56,
|
| 174 |
+
"learning_rate": 0.0,
|
| 175 |
+
"loss": 0.371,
|
| 176 |
+
"step": 260
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.62,
|
| 180 |
+
"learning_rate": 0.0,
|
| 181 |
+
"loss": 0.3686,
|
| 182 |
+
"step": 270
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.68,
|
| 186 |
+
"learning_rate": 0.0,
|
| 187 |
+
"loss": 0.3774,
|
| 188 |
+
"step": 280
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 1.74,
|
| 192 |
+
"learning_rate": 0.0,
|
| 193 |
+
"loss": 0.3703,
|
| 194 |
+
"step": 290
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.8,
|
| 198 |
+
"learning_rate": 0.0,
|
| 199 |
+
"loss": 0.3731,
|
| 200 |
+
"step": 300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.86,
|
| 204 |
+
"learning_rate": 0.0,
|
| 205 |
+
"loss": 0.3763,
|
| 206 |
+
"step": 310
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.92,
|
| 210 |
+
"learning_rate": 0.0,
|
| 211 |
+
"loss": 0.3712,
|
| 212 |
+
"step": 320
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.98,
|
| 216 |
+
"learning_rate": 0.0,
|
| 217 |
+
"loss": 0.3762,
|
| 218 |
+
"step": 330
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.0,
|
| 222 |
+
"eval_accuracy": 0.984984984984985,
|
| 223 |
+
"eval_f1": 0.9849327674451294,
|
| 224 |
+
"eval_loss": 0.2975810170173645,
|
| 225 |
+
"eval_precision": 0.9865442976554087,
|
| 226 |
+
"eval_recall": 0.984984984984985,
|
| 227 |
+
"eval_runtime": 3.6994,
|
| 228 |
+
"eval_samples_per_second": 180.03,
|
| 229 |
+
"eval_steps_per_second": 5.677,
|
| 230 |
+
"step": 334
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.04,
|
| 234 |
+
"learning_rate": 0.0,
|
| 235 |
+
"loss": 0.3601,
|
| 236 |
+
"step": 340
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.1,
|
| 240 |
+
"learning_rate": 0.0,
|
| 241 |
+
"loss": 0.3885,
|
| 242 |
+
"step": 350
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 2.16,
|
| 246 |
+
"learning_rate": 0.0,
|
| 247 |
+
"loss": 0.3864,
|
| 248 |
+
"step": 360
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.22,
|
| 252 |
+
"learning_rate": 0.0,
|
| 253 |
+
"loss": 0.3701,
|
| 254 |
+
"step": 370
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 2.28,
|
| 258 |
+
"learning_rate": 0.0,
|
| 259 |
+
"loss": 0.3744,
|
| 260 |
+
"step": 380
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 2.34,
|
| 264 |
+
"learning_rate": 0.0,
|
| 265 |
+
"loss": 0.3791,
|
| 266 |
+
"step": 390
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.4,
|
| 270 |
+
"learning_rate": 0.0,
|
| 271 |
+
"loss": 0.3572,
|
| 272 |
+
"step": 400
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.46,
|
| 276 |
+
"learning_rate": 0.0,
|
| 277 |
+
"loss": 0.3809,
|
| 278 |
+
"step": 410
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.51,
|
| 282 |
+
"learning_rate": 0.0,
|
| 283 |
+
"loss": 0.3796,
|
| 284 |
+
"step": 420
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 2.57,
|
| 288 |
+
"learning_rate": 0.0,
|
| 289 |
+
"loss": 0.3517,
|
| 290 |
+
"step": 430
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.63,
|
| 294 |
+
"learning_rate": 0.0,
|
| 295 |
+
"loss": 0.3723,
|
| 296 |
+
"step": 440
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.69,
|
| 300 |
+
"learning_rate": 0.0,
|
| 301 |
+
"loss": 0.3895,
|
| 302 |
+
"step": 450
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.75,
|
| 306 |
+
"learning_rate": 0.0,
|
| 307 |
+
"loss": 0.3786,
|
| 308 |
+
"step": 460
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 2.81,
|
| 312 |
+
"learning_rate": 0.0,
|
| 313 |
+
"loss": 0.3844,
|
| 314 |
+
"step": 470
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.87,
|
| 318 |
+
"learning_rate": 0.0,
|
| 319 |
+
"loss": 0.3575,
|
| 320 |
+
"step": 480
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.93,
|
| 324 |
+
"learning_rate": 0.0,
|
| 325 |
+
"loss": 0.3547,
|
| 326 |
+
"step": 490
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.99,
|
| 330 |
+
"learning_rate": 0.0,
|
| 331 |
+
"loss": 0.3783,
|
| 332 |
+
"step": 500
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 3.0,
|
| 336 |
+
"eval_accuracy": 0.984984984984985,
|
| 337 |
+
"eval_f1": 0.9849327674451294,
|
| 338 |
+
"eval_loss": 0.2975810170173645,
|
| 339 |
+
"eval_precision": 0.9865442976554087,
|
| 340 |
+
"eval_recall": 0.984984984984985,
|
| 341 |
+
"eval_runtime": 3.693,
|
| 342 |
+
"eval_samples_per_second": 180.342,
|
| 343 |
+
"eval_steps_per_second": 5.686,
|
| 344 |
+
"step": 501
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 3.0,
|
| 348 |
+
"step": 501,
|
| 349 |
+
"total_flos": 2118227382190080.0,
|
| 350 |
+
"train_loss": 0.374498660454969,
|
| 351 |
+
"train_runtime": 306.4673,
|
| 352 |
+
"train_samples_per_second": 52.156,
|
| 353 |
+
"train_steps_per_second": 1.635
|
| 354 |
+
}
|
| 355 |
+
],
|
| 356 |
+
"max_steps": 501,
|
| 357 |
+
"num_train_epochs": 3,
|
| 358 |
+
"total_flos": 2118227382190080.0,
|
| 359 |
+
"trial_name": null,
|
| 360 |
+
"trial_params": null
|
| 361 |
+
}
|