End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.6819923371647509,
|
| 4 |
+
"eval_loss": 0.8754438161849976,
|
| 5 |
+
"eval_runtime": 25.581,
|
| 6 |
+
"eval_samples_per_second": 112.232,
|
| 7 |
+
"eval_steps_per_second": 3.518,
|
| 8 |
+
"total_flos": 6.004415924974301e+18,
|
| 9 |
+
"train_loss": 1.159771179602091,
|
| 10 |
+
"train_runtime": 1779.2361,
|
| 11 |
+
"train_samples_per_second": 43.566,
|
| 12 |
+
"train_steps_per_second": 0.341
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.6819923371647509,
|
| 4 |
+
"eval_loss": 0.8754438161849976,
|
| 5 |
+
"eval_runtime": 25.581,
|
| 6 |
+
"eval_samples_per_second": 112.232,
|
| 7 |
+
"eval_steps_per_second": 3.518
|
| 8 |
+
}
|
runs/Jan07_22-09-54_1f4fb9c623db/events.out.tfevents.1673131316.1f4fb9c623db.23.5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:034457ecea503cf0990b26b3c6478cafd7be70256edd43295d306a89c3e7fffb
|
| 3 |
+
size 363
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6.004415924974301e+18,
|
| 4 |
+
"train_loss": 1.159771179602091,
|
| 5 |
+
"train_runtime": 1779.2361,
|
| 6 |
+
"train_samples_per_second": 43.566,
|
| 7 |
+
"train_steps_per_second": 0.341
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,412 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.6819923371647509,
|
| 3 |
+
"best_model_checkpoint": "beit-base-patch16-224-pt22k-ft22k-finetuned-FER2013/checkpoint-606",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"global_step": 606,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.05,
|
| 12 |
+
"learning_rate": 8.196721311475409e-06,
|
| 13 |
+
"loss": 1.9392,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.1,
|
| 18 |
+
"learning_rate": 1.6393442622950818e-05,
|
| 19 |
+
"loss": 1.7671,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.15,
|
| 24 |
+
"learning_rate": 2.459016393442623e-05,
|
| 25 |
+
"loss": 1.6727,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.2,
|
| 30 |
+
"learning_rate": 3.2786885245901635e-05,
|
| 31 |
+
"loss": 1.4688,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.25,
|
| 36 |
+
"learning_rate": 4.098360655737705e-05,
|
| 37 |
+
"loss": 1.4389,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.3,
|
| 42 |
+
"learning_rate": 4.918032786885246e-05,
|
| 43 |
+
"loss": 1.3738,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.35,
|
| 48 |
+
"learning_rate": 4.917431192660551e-05,
|
| 49 |
+
"loss": 1.3653,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.4,
|
| 54 |
+
"learning_rate": 4.8256880733944956e-05,
|
| 55 |
+
"loss": 1.2999,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.45,
|
| 60 |
+
"learning_rate": 4.733944954128441e-05,
|
| 61 |
+
"loss": 1.301,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.5,
|
| 66 |
+
"learning_rate": 4.642201834862386e-05,
|
| 67 |
+
"loss": 1.3416,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.54,
|
| 72 |
+
"learning_rate": 4.5504587155963305e-05,
|
| 73 |
+
"loss": 1.2908,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.59,
|
| 78 |
+
"learning_rate": 4.458715596330276e-05,
|
| 79 |
+
"loss": 1.2205,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.64,
|
| 84 |
+
"learning_rate": 4.366972477064221e-05,
|
| 85 |
+
"loss": 1.2779,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.69,
|
| 90 |
+
"learning_rate": 4.2752293577981654e-05,
|
| 91 |
+
"loss": 1.2563,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.74,
|
| 96 |
+
"learning_rate": 4.1834862385321106e-05,
|
| 97 |
+
"loss": 1.2122,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.79,
|
| 102 |
+
"learning_rate": 4.091743119266056e-05,
|
| 103 |
+
"loss": 1.2289,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.84,
|
| 108 |
+
"learning_rate": 4e-05,
|
| 109 |
+
"loss": 1.1613,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.89,
|
| 114 |
+
"learning_rate": 3.9082568807339455e-05,
|
| 115 |
+
"loss": 1.1646,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.94,
|
| 120 |
+
"learning_rate": 3.81651376146789e-05,
|
| 121 |
+
"loss": 1.2029,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.99,
|
| 126 |
+
"learning_rate": 3.724770642201835e-05,
|
| 127 |
+
"loss": 1.1701,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.0,
|
| 132 |
+
"eval_accuracy": 0.6179031696273075,
|
| 133 |
+
"eval_loss": 1.0163276195526123,
|
| 134 |
+
"eval_runtime": 25.3453,
|
| 135 |
+
"eval_samples_per_second": 113.276,
|
| 136 |
+
"eval_steps_per_second": 3.551,
|
| 137 |
+
"step": 202
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.04,
|
| 141 |
+
"learning_rate": 3.6330275229357804e-05,
|
| 142 |
+
"loss": 1.1281,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 1.09,
|
| 147 |
+
"learning_rate": 3.541284403669725e-05,
|
| 148 |
+
"loss": 1.1595,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.14,
|
| 153 |
+
"learning_rate": 3.44954128440367e-05,
|
| 154 |
+
"loss": 1.1224,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.19,
|
| 159 |
+
"learning_rate": 3.3577981651376154e-05,
|
| 160 |
+
"loss": 1.0808,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.24,
|
| 165 |
+
"learning_rate": 3.26605504587156e-05,
|
| 166 |
+
"loss": 1.1312,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.29,
|
| 171 |
+
"learning_rate": 3.174311926605505e-05,
|
| 172 |
+
"loss": 1.1423,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.34,
|
| 177 |
+
"learning_rate": 3.0825688073394496e-05,
|
| 178 |
+
"loss": 1.1882,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.39,
|
| 183 |
+
"learning_rate": 2.9908256880733948e-05,
|
| 184 |
+
"loss": 1.1086,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.44,
|
| 189 |
+
"learning_rate": 2.8990825688073397e-05,
|
| 190 |
+
"loss": 1.0772,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.49,
|
| 195 |
+
"learning_rate": 2.807339449541285e-05,
|
| 196 |
+
"loss": 1.1303,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.53,
|
| 201 |
+
"learning_rate": 2.7155963302752297e-05,
|
| 202 |
+
"loss": 1.1025,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.58,
|
| 207 |
+
"learning_rate": 2.6238532110091746e-05,
|
| 208 |
+
"loss": 1.1117,
|
| 209 |
+
"step": 320
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.63,
|
| 213 |
+
"learning_rate": 2.5321100917431194e-05,
|
| 214 |
+
"loss": 1.0372,
|
| 215 |
+
"step": 330
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.68,
|
| 219 |
+
"learning_rate": 2.4403669724770646e-05,
|
| 220 |
+
"loss": 1.0228,
|
| 221 |
+
"step": 340
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.73,
|
| 225 |
+
"learning_rate": 2.3486238532110095e-05,
|
| 226 |
+
"loss": 1.0936,
|
| 227 |
+
"step": 350
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.78,
|
| 231 |
+
"learning_rate": 2.2568807339449544e-05,
|
| 232 |
+
"loss": 1.0889,
|
| 233 |
+
"step": 360
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.83,
|
| 237 |
+
"learning_rate": 2.1651376146788992e-05,
|
| 238 |
+
"loss": 1.0765,
|
| 239 |
+
"step": 370
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.88,
|
| 243 |
+
"learning_rate": 2.0733944954128444e-05,
|
| 244 |
+
"loss": 1.0954,
|
| 245 |
+
"step": 380
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.93,
|
| 249 |
+
"learning_rate": 1.9816513761467893e-05,
|
| 250 |
+
"loss": 1.1196,
|
| 251 |
+
"step": 390
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.98,
|
| 255 |
+
"learning_rate": 1.889908256880734e-05,
|
| 256 |
+
"loss": 1.0447,
|
| 257 |
+
"step": 400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 2.0,
|
| 261 |
+
"eval_accuracy": 0.6569139672587948,
|
| 262 |
+
"eval_loss": 0.9236605763435364,
|
| 263 |
+
"eval_runtime": 25.7906,
|
| 264 |
+
"eval_samples_per_second": 111.32,
|
| 265 |
+
"eval_steps_per_second": 3.49,
|
| 266 |
+
"step": 404
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 2.03,
|
| 270 |
+
"learning_rate": 1.798165137614679e-05,
|
| 271 |
+
"loss": 1.0341,
|
| 272 |
+
"step": 410
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 2.08,
|
| 276 |
+
"learning_rate": 1.7064220183486242e-05,
|
| 277 |
+
"loss": 0.9925,
|
| 278 |
+
"step": 420
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 2.13,
|
| 282 |
+
"learning_rate": 1.614678899082569e-05,
|
| 283 |
+
"loss": 1.0571,
|
| 284 |
+
"step": 430
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 2.18,
|
| 288 |
+
"learning_rate": 1.5229357798165139e-05,
|
| 289 |
+
"loss": 1.0625,
|
| 290 |
+
"step": 440
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.23,
|
| 294 |
+
"learning_rate": 1.431192660550459e-05,
|
| 295 |
+
"loss": 1.0117,
|
| 296 |
+
"step": 450
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.28,
|
| 300 |
+
"learning_rate": 1.3394495412844038e-05,
|
| 301 |
+
"loss": 0.9992,
|
| 302 |
+
"step": 460
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.33,
|
| 306 |
+
"learning_rate": 1.2477064220183488e-05,
|
| 307 |
+
"loss": 1.0723,
|
| 308 |
+
"step": 470
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 2.38,
|
| 312 |
+
"learning_rate": 1.1559633027522937e-05,
|
| 313 |
+
"loss": 1.0284,
|
| 314 |
+
"step": 480
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.43,
|
| 318 |
+
"learning_rate": 1.0642201834862387e-05,
|
| 319 |
+
"loss": 0.9993,
|
| 320 |
+
"step": 490
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.48,
|
| 324 |
+
"learning_rate": 9.724770642201836e-06,
|
| 325 |
+
"loss": 1.023,
|
| 326 |
+
"step": 500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.52,
|
| 330 |
+
"learning_rate": 8.807339449541286e-06,
|
| 331 |
+
"loss": 1.0313,
|
| 332 |
+
"step": 510
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.57,
|
| 336 |
+
"learning_rate": 7.889908256880735e-06,
|
| 337 |
+
"loss": 1.0584,
|
| 338 |
+
"step": 520
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.62,
|
| 342 |
+
"learning_rate": 6.972477064220184e-06,
|
| 343 |
+
"loss": 1.003,
|
| 344 |
+
"step": 530
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.67,
|
| 348 |
+
"learning_rate": 6.0550458715596335e-06,
|
| 349 |
+
"loss": 1.0541,
|
| 350 |
+
"step": 540
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.72,
|
| 354 |
+
"learning_rate": 5.137614678899083e-06,
|
| 355 |
+
"loss": 1.0189,
|
| 356 |
+
"step": 550
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.77,
|
| 360 |
+
"learning_rate": 4.220183486238532e-06,
|
| 361 |
+
"loss": 0.9714,
|
| 362 |
+
"step": 560
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.82,
|
| 366 |
+
"learning_rate": 3.302752293577982e-06,
|
| 367 |
+
"loss": 1.0046,
|
| 368 |
+
"step": 570
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.87,
|
| 372 |
+
"learning_rate": 2.3853211009174313e-06,
|
| 373 |
+
"loss": 1.0082,
|
| 374 |
+
"step": 580
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.92,
|
| 378 |
+
"learning_rate": 1.467889908256881e-06,
|
| 379 |
+
"loss": 1.0499,
|
| 380 |
+
"step": 590
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.97,
|
| 384 |
+
"learning_rate": 5.504587155963304e-07,
|
| 385 |
+
"loss": 0.9712,
|
| 386 |
+
"step": 600
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 3.0,
|
| 390 |
+
"eval_accuracy": 0.6819923371647509,
|
| 391 |
+
"eval_loss": 0.8754438161849976,
|
| 392 |
+
"eval_runtime": 25.5354,
|
| 393 |
+
"eval_samples_per_second": 112.432,
|
| 394 |
+
"eval_steps_per_second": 3.525,
|
| 395 |
+
"step": 606
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 3.0,
|
| 399 |
+
"step": 606,
|
| 400 |
+
"total_flos": 6.004415924974301e+18,
|
| 401 |
+
"train_loss": 1.159771179602091,
|
| 402 |
+
"train_runtime": 1779.2361,
|
| 403 |
+
"train_samples_per_second": 43.566,
|
| 404 |
+
"train_steps_per_second": 0.341
|
| 405 |
+
}
|
| 406 |
+
],
|
| 407 |
+
"max_steps": 606,
|
| 408 |
+
"num_train_epochs": 3,
|
| 409 |
+
"total_flos": 6.004415924974301e+18,
|
| 410 |
+
"trial_name": null,
|
| 411 |
+
"trial_params": null
|
| 412 |
+
}
|