End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 15.0,
|
| 3 |
+
"eval_accuracy": 0.9003656506538391,
|
| 4 |
+
"eval_loss": 0.5111713409423828,
|
| 5 |
+
"eval_runtime": 2.6828,
|
| 6 |
+
"eval_samples": 1094,
|
| 7 |
+
"eval_samples_per_second": 407.786,
|
| 8 |
+
"eval_steps_per_second": 51.066,
|
| 9 |
+
"train_loss": 0.8604840965945312,
|
| 10 |
+
"train_runtime": 2276.2445,
|
| 11 |
+
"train_samples": 10556,
|
| 12 |
+
"train_samples_per_second": 69.562,
|
| 13 |
+
"train_steps_per_second": 8.699
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 15.0,
|
| 3 |
+
"eval_accuracy": 0.9003656506538391,
|
| 4 |
+
"eval_loss": 0.5111713409423828,
|
| 5 |
+
"eval_runtime": 2.6828,
|
| 6 |
+
"eval_samples": 1094,
|
| 7 |
+
"eval_samples_per_second": 407.786,
|
| 8 |
+
"eval_steps_per_second": 51.066
|
| 9 |
+
}
|
runs/Dec07_13-07-11_jbuhmann/events.out.tfevents.1638882455.jbuhmann.31503.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ed14e2390c3d7741a52f10c299b5906f1435fe1189e573ec7d6f4806dd9bfa13
|
| 3 |
+
size 27721
|
runs/Dec07_13-07-11_jbuhmann/events.out.tfevents.1638884736.jbuhmann.31503.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7dd15b4ac152154e9d94795fa4236a969c333929258f61f2f58a90ad33994e6a
|
| 3 |
+
size 369
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 15.0,
|
| 3 |
+
"train_loss": 0.8604840965945312,
|
| 4 |
+
"train_runtime": 2276.2445,
|
| 5 |
+
"train_samples": 10556,
|
| 6 |
+
"train_samples_per_second": 69.562,
|
| 7 |
+
"train_steps_per_second": 8.699
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,394 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 15.0,
|
| 5 |
+
"global_step": 19800,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.38,
|
| 12 |
+
"learning_rate": 9.747474747474748e-06,
|
| 13 |
+
"loss": 4.7663,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.76,
|
| 18 |
+
"learning_rate": 9.494949494949497e-06,
|
| 19 |
+
"loss": 4.1505,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.0,
|
| 24 |
+
"eval_accuracy": 0.37934187054634094,
|
| 25 |
+
"eval_loss": 3.329326629638672,
|
| 26 |
+
"eval_runtime": 3.3016,
|
| 27 |
+
"eval_samples_per_second": 331.353,
|
| 28 |
+
"eval_steps_per_second": 41.495,
|
| 29 |
+
"step": 1320
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 1.14,
|
| 33 |
+
"learning_rate": 9.242424242424244e-06,
|
| 34 |
+
"loss": 3.6225,
|
| 35 |
+
"step": 1500
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 1.52,
|
| 39 |
+
"learning_rate": 8.98989898989899e-06,
|
| 40 |
+
"loss": 3.0767,
|
| 41 |
+
"step": 2000
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 1.89,
|
| 45 |
+
"learning_rate": 8.737373737373738e-06,
|
| 46 |
+
"loss": 2.7333,
|
| 47 |
+
"step": 2500
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 2.0,
|
| 51 |
+
"eval_accuracy": 0.6133455038070679,
|
| 52 |
+
"eval_loss": 2.2294859886169434,
|
| 53 |
+
"eval_runtime": 2.0812,
|
| 54 |
+
"eval_samples_per_second": 525.664,
|
| 55 |
+
"eval_steps_per_second": 65.828,
|
| 56 |
+
"step": 2640
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 2.27,
|
| 60 |
+
"learning_rate": 8.484848484848486e-06,
|
| 61 |
+
"loss": 2.3064,
|
| 62 |
+
"step": 3000
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 2.65,
|
| 66 |
+
"learning_rate": 8.232323232323233e-06,
|
| 67 |
+
"loss": 2.0189,
|
| 68 |
+
"step": 3500
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 3.0,
|
| 72 |
+
"eval_accuracy": 0.7586837410926819,
|
| 73 |
+
"eval_loss": 1.5133757591247559,
|
| 74 |
+
"eval_runtime": 2.0858,
|
| 75 |
+
"eval_samples_per_second": 524.509,
|
| 76 |
+
"eval_steps_per_second": 65.683,
|
| 77 |
+
"step": 3960
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 3.03,
|
| 81 |
+
"learning_rate": 7.97979797979798e-06,
|
| 82 |
+
"loss": 1.7059,
|
| 83 |
+
"step": 4000
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 3.41,
|
| 87 |
+
"learning_rate": 7.727272727272727e-06,
|
| 88 |
+
"loss": 1.4008,
|
| 89 |
+
"step": 4500
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 3.79,
|
| 93 |
+
"learning_rate": 7.474747474747476e-06,
|
| 94 |
+
"loss": 1.2504,
|
| 95 |
+
"step": 5000
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 4.0,
|
| 99 |
+
"eval_accuracy": 0.8281535506248474,
|
| 100 |
+
"eval_loss": 1.0764867067337036,
|
| 101 |
+
"eval_runtime": 2.3812,
|
| 102 |
+
"eval_samples_per_second": 459.44,
|
| 103 |
+
"eval_steps_per_second": 57.535,
|
| 104 |
+
"step": 5280
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 4.17,
|
| 108 |
+
"learning_rate": 7.222222222222223e-06,
|
| 109 |
+
"loss": 1.0472,
|
| 110 |
+
"step": 5500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 4.55,
|
| 114 |
+
"learning_rate": 6.969696969696971e-06,
|
| 115 |
+
"loss": 0.8521,
|
| 116 |
+
"step": 6000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 4.92,
|
| 120 |
+
"learning_rate": 6.717171717171718e-06,
|
| 121 |
+
"loss": 0.7733,
|
| 122 |
+
"step": 6500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 5.0,
|
| 126 |
+
"eval_accuracy": 0.8628884553909302,
|
| 127 |
+
"eval_loss": 0.7936682105064392,
|
| 128 |
+
"eval_runtime": 2.4487,
|
| 129 |
+
"eval_samples_per_second": 446.77,
|
| 130 |
+
"eval_steps_per_second": 55.948,
|
| 131 |
+
"step": 6600
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 5.3,
|
| 135 |
+
"learning_rate": 6.464646464646466e-06,
|
| 136 |
+
"loss": 0.6059,
|
| 137 |
+
"step": 7000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 5.68,
|
| 141 |
+
"learning_rate": 6.212121212121213e-06,
|
| 142 |
+
"loss": 0.5217,
|
| 143 |
+
"step": 7500
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 6.0,
|
| 147 |
+
"eval_accuracy": 0.8784278035163879,
|
| 148 |
+
"eval_loss": 0.643801212310791,
|
| 149 |
+
"eval_runtime": 2.8497,
|
| 150 |
+
"eval_samples_per_second": 383.9,
|
| 151 |
+
"eval_steps_per_second": 48.075,
|
| 152 |
+
"step": 7920
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 6.06,
|
| 156 |
+
"learning_rate": 5.95959595959596e-06,
|
| 157 |
+
"loss": 0.4521,
|
| 158 |
+
"step": 8000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 6.44,
|
| 162 |
+
"learning_rate": 5.7070707070707075e-06,
|
| 163 |
+
"loss": 0.3498,
|
| 164 |
+
"step": 8500
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 6.82,
|
| 168 |
+
"learning_rate": 5.4545454545454545e-06,
|
| 169 |
+
"loss": 0.3148,
|
| 170 |
+
"step": 9000
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 7.0,
|
| 174 |
+
"eval_accuracy": 0.8857403993606567,
|
| 175 |
+
"eval_loss": 0.5733386874198914,
|
| 176 |
+
"eval_runtime": 2.4195,
|
| 177 |
+
"eval_samples_per_second": 452.164,
|
| 178 |
+
"eval_steps_per_second": 56.624,
|
| 179 |
+
"step": 9240
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 7.2,
|
| 183 |
+
"learning_rate": 5.202020202020202e-06,
|
| 184 |
+
"loss": 0.2811,
|
| 185 |
+
"step": 9500
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 7.58,
|
| 189 |
+
"learning_rate": 4.94949494949495e-06,
|
| 190 |
+
"loss": 0.2148,
|
| 191 |
+
"step": 10000
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 7.95,
|
| 195 |
+
"learning_rate": 4.696969696969698e-06,
|
| 196 |
+
"loss": 0.2067,
|
| 197 |
+
"step": 10500
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 8.0,
|
| 201 |
+
"eval_accuracy": 0.8912248611450195,
|
| 202 |
+
"eval_loss": 0.536188542842865,
|
| 203 |
+
"eval_runtime": 2.4481,
|
| 204 |
+
"eval_samples_per_second": 446.874,
|
| 205 |
+
"eval_steps_per_second": 55.961,
|
| 206 |
+
"step": 10560
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 8.33,
|
| 210 |
+
"learning_rate": 4.444444444444444e-06,
|
| 211 |
+
"loss": 0.1633,
|
| 212 |
+
"step": 11000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 8.71,
|
| 216 |
+
"learning_rate": 4.191919191919192e-06,
|
| 217 |
+
"loss": 0.1507,
|
| 218 |
+
"step": 11500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 9.0,
|
| 222 |
+
"eval_accuracy": 0.8903107643127441,
|
| 223 |
+
"eval_loss": 0.5098453164100647,
|
| 224 |
+
"eval_runtime": 2.5274,
|
| 225 |
+
"eval_samples_per_second": 432.848,
|
| 226 |
+
"eval_steps_per_second": 54.205,
|
| 227 |
+
"step": 11880
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 9.09,
|
| 231 |
+
"learning_rate": 3.93939393939394e-06,
|
| 232 |
+
"loss": 0.1385,
|
| 233 |
+
"step": 12000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 9.47,
|
| 237 |
+
"learning_rate": 3.686868686868687e-06,
|
| 238 |
+
"loss": 0.1079,
|
| 239 |
+
"step": 12500
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 9.85,
|
| 243 |
+
"learning_rate": 3.4343434343434347e-06,
|
| 244 |
+
"loss": 0.1024,
|
| 245 |
+
"step": 13000
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 10.0,
|
| 249 |
+
"eval_accuracy": 0.8976234197616577,
|
| 250 |
+
"eval_loss": 0.5077915191650391,
|
| 251 |
+
"eval_runtime": 2.4776,
|
| 252 |
+
"eval_samples_per_second": 441.549,
|
| 253 |
+
"eval_steps_per_second": 55.295,
|
| 254 |
+
"step": 13200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 10.23,
|
| 258 |
+
"learning_rate": 3.181818181818182e-06,
|
| 259 |
+
"loss": 0.083,
|
| 260 |
+
"step": 13500
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 10.61,
|
| 264 |
+
"learning_rate": 2.9292929292929295e-06,
|
| 265 |
+
"loss": 0.0784,
|
| 266 |
+
"step": 14000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 10.98,
|
| 270 |
+
"learning_rate": 2.676767676767677e-06,
|
| 271 |
+
"loss": 0.0837,
|
| 272 |
+
"step": 14500
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 11.0,
|
| 276 |
+
"eval_accuracy": 0.8967093229293823,
|
| 277 |
+
"eval_loss": 0.5054203271865845,
|
| 278 |
+
"eval_runtime": 2.4447,
|
| 279 |
+
"eval_samples_per_second": 447.49,
|
| 280 |
+
"eval_steps_per_second": 56.039,
|
| 281 |
+
"step": 14520
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 11.36,
|
| 285 |
+
"learning_rate": 2.4242424242424244e-06,
|
| 286 |
+
"loss": 0.0587,
|
| 287 |
+
"step": 15000
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 11.74,
|
| 291 |
+
"learning_rate": 2.171717171717172e-06,
|
| 292 |
+
"loss": 0.0608,
|
| 293 |
+
"step": 15500
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 12.0,
|
| 297 |
+
"eval_accuracy": 0.8957952260971069,
|
| 298 |
+
"eval_loss": 0.5061630606651306,
|
| 299 |
+
"eval_runtime": 2.4836,
|
| 300 |
+
"eval_samples_per_second": 440.493,
|
| 301 |
+
"eval_steps_per_second": 55.162,
|
| 302 |
+
"step": 15840
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 12.12,
|
| 306 |
+
"learning_rate": 1.9191919191919192e-06,
|
| 307 |
+
"loss": 0.0601,
|
| 308 |
+
"step": 16000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 12.5,
|
| 312 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 313 |
+
"loss": 0.0561,
|
| 314 |
+
"step": 16500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 12.88,
|
| 318 |
+
"learning_rate": 1.4141414141414143e-06,
|
| 319 |
+
"loss": 0.0426,
|
| 320 |
+
"step": 17000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 13.0,
|
| 324 |
+
"eval_accuracy": 0.9012796878814697,
|
| 325 |
+
"eval_loss": 0.5072430968284607,
|
| 326 |
+
"eval_runtime": 2.5695,
|
| 327 |
+
"eval_samples_per_second": 425.758,
|
| 328 |
+
"eval_steps_per_second": 53.317,
|
| 329 |
+
"step": 17160
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 13.26,
|
| 333 |
+
"learning_rate": 1.1616161616161617e-06,
|
| 334 |
+
"loss": 0.0493,
|
| 335 |
+
"step": 17500
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 13.64,
|
| 339 |
+
"learning_rate": 9.090909090909091e-07,
|
| 340 |
+
"loss": 0.0374,
|
| 341 |
+
"step": 18000
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 14.0,
|
| 345 |
+
"eval_accuracy": 0.9040219187736511,
|
| 346 |
+
"eval_loss": 0.5110482573509216,
|
| 347 |
+
"eval_runtime": 2.4737,
|
| 348 |
+
"eval_samples_per_second": 442.248,
|
| 349 |
+
"eval_steps_per_second": 55.382,
|
| 350 |
+
"step": 18480
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 14.02,
|
| 354 |
+
"learning_rate": 6.565656565656567e-07,
|
| 355 |
+
"loss": 0.0491,
|
| 356 |
+
"step": 18500
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 14.39,
|
| 360 |
+
"learning_rate": 4.040404040404041e-07,
|
| 361 |
+
"loss": 0.041,
|
| 362 |
+
"step": 19000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 14.77,
|
| 366 |
+
"learning_rate": 1.5151515151515152e-07,
|
| 367 |
+
"loss": 0.0346,
|
| 368 |
+
"step": 19500
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 15.0,
|
| 372 |
+
"eval_accuracy": 0.9003656506538391,
|
| 373 |
+
"eval_loss": 0.5111713409423828,
|
| 374 |
+
"eval_runtime": 2.5194,
|
| 375 |
+
"eval_samples_per_second": 434.227,
|
| 376 |
+
"eval_steps_per_second": 54.378,
|
| 377 |
+
"step": 19800
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 15.0,
|
| 381 |
+
"step": 19800,
|
| 382 |
+
"total_flos": 5215995096399360.0,
|
| 383 |
+
"train_loss": 0.8604840965945312,
|
| 384 |
+
"train_runtime": 2276.2445,
|
| 385 |
+
"train_samples_per_second": 69.562,
|
| 386 |
+
"train_steps_per_second": 8.699
|
| 387 |
+
}
|
| 388 |
+
],
|
| 389 |
+
"max_steps": 19800,
|
| 390 |
+
"num_train_epochs": 15,
|
| 391 |
+
"total_flos": 5215995096399360.0,
|
| 392 |
+
"trial_name": null,
|
| 393 |
+
"trial_params": null
|
| 394 |
+
}
|