End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.6851490665923655,
|
| 4 |
+
"eval_loss": 0.8512730002403259,
|
| 5 |
+
"eval_runtime": 65.2334,
|
| 6 |
+
"eval_samples_per_second": 110.036,
|
| 7 |
+
"eval_steps_per_second": 3.449,
|
| 8 |
+
"total_flos": 6.668732964123095e+18,
|
| 9 |
+
"train_loss": 1.145610089103381,
|
| 10 |
+
"train_runtime": 2105.9038,
|
| 11 |
+
"train_samples_per_second": 40.898,
|
| 12 |
+
"train_steps_per_second": 0.319
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.6851490665923655,
|
| 4 |
+
"eval_loss": 0.8512730002403259,
|
| 5 |
+
"eval_runtime": 65.2334,
|
| 6 |
+
"eval_samples_per_second": 110.036,
|
| 7 |
+
"eval_steps_per_second": 3.449
|
| 8 |
+
}
|
runs/Jan09_11-29-02_f950aa079631/events.out.tfevents.1673266029.f950aa079631.23.5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f40e97abcfd025a6f777ebf4cb49b5853f2d97c7cd7c254d4e5fe3b38e5c6b88
|
| 3 |
+
size 363
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6.668732964123095e+18,
|
| 4 |
+
"train_loss": 1.145610089103381,
|
| 5 |
+
"train_runtime": 2105.9038,
|
| 6 |
+
"train_samples_per_second": 40.898,
|
| 7 |
+
"train_steps_per_second": 0.319
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,454 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.6851490665923655,
|
| 3 |
+
"best_model_checkpoint": "beit-base-patch16-224-pt22k-ft22k-finetuned-FER2013-8e-05/checkpoint-672",
|
| 4 |
+
"epoch": 2.997772828507795,
|
| 5 |
+
"global_step": 672,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.04,
|
| 12 |
+
"learning_rate": 1.1764705882352942e-05,
|
| 13 |
+
"loss": 1.9257,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.09,
|
| 18 |
+
"learning_rate": 2.3529411764705884e-05,
|
| 19 |
+
"loss": 1.7335,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.13,
|
| 24 |
+
"learning_rate": 3.529411764705883e-05,
|
| 25 |
+
"loss": 1.5648,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.18,
|
| 30 |
+
"learning_rate": 4.705882352941177e-05,
|
| 31 |
+
"loss": 1.4534,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.22,
|
| 36 |
+
"learning_rate": 5.8823529411764714e-05,
|
| 37 |
+
"loss": 1.4039,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.27,
|
| 42 |
+
"learning_rate": 7.058823529411765e-05,
|
| 43 |
+
"loss": 1.3592,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.31,
|
| 48 |
+
"learning_rate": 7.973509933774835e-05,
|
| 49 |
+
"loss": 1.3827,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.36,
|
| 54 |
+
"learning_rate": 7.841059602649007e-05,
|
| 55 |
+
"loss": 1.2943,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.4,
|
| 60 |
+
"learning_rate": 7.70860927152318e-05,
|
| 61 |
+
"loss": 1.2908,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.45,
|
| 66 |
+
"learning_rate": 7.576158940397351e-05,
|
| 67 |
+
"loss": 1.3016,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.49,
|
| 72 |
+
"learning_rate": 7.443708609271524e-05,
|
| 73 |
+
"loss": 1.2757,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.53,
|
| 78 |
+
"learning_rate": 7.311258278145696e-05,
|
| 79 |
+
"loss": 1.2541,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.58,
|
| 84 |
+
"learning_rate": 7.178807947019869e-05,
|
| 85 |
+
"loss": 1.2023,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.62,
|
| 90 |
+
"learning_rate": 7.04635761589404e-05,
|
| 91 |
+
"loss": 1.3072,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.67,
|
| 96 |
+
"learning_rate": 6.913907284768213e-05,
|
| 97 |
+
"loss": 1.2497,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.71,
|
| 102 |
+
"learning_rate": 6.781456953642385e-05,
|
| 103 |
+
"loss": 1.2927,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.76,
|
| 108 |
+
"learning_rate": 6.649006622516556e-05,
|
| 109 |
+
"loss": 1.2041,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.8,
|
| 114 |
+
"learning_rate": 6.516556291390729e-05,
|
| 115 |
+
"loss": 1.199,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.85,
|
| 120 |
+
"learning_rate": 6.3841059602649e-05,
|
| 121 |
+
"loss": 1.2165,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.89,
|
| 126 |
+
"learning_rate": 6.251655629139073e-05,
|
| 127 |
+
"loss": 1.1911,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.94,
|
| 132 |
+
"learning_rate": 6.119205298013245e-05,
|
| 133 |
+
"loss": 1.1666,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.98,
|
| 138 |
+
"learning_rate": 5.986754966887418e-05,
|
| 139 |
+
"loss": 1.1538,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.0,
|
| 144 |
+
"eval_accuracy": 0.6173028698801895,
|
| 145 |
+
"eval_loss": 1.0146794319152832,
|
| 146 |
+
"eval_runtime": 61.7913,
|
| 147 |
+
"eval_samples_per_second": 116.165,
|
| 148 |
+
"eval_steps_per_second": 3.641,
|
| 149 |
+
"step": 224
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.03,
|
| 153 |
+
"learning_rate": 5.85430463576159e-05,
|
| 154 |
+
"loss": 1.2208,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.07,
|
| 159 |
+
"learning_rate": 5.721854304635762e-05,
|
| 160 |
+
"loss": 1.1085,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.12,
|
| 165 |
+
"learning_rate": 5.589403973509934e-05,
|
| 166 |
+
"loss": 1.1218,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.16,
|
| 171 |
+
"learning_rate": 5.456953642384107e-05,
|
| 172 |
+
"loss": 1.1537,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 1.2,
|
| 177 |
+
"learning_rate": 5.324503311258278e-05,
|
| 178 |
+
"loss": 1.1485,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.25,
|
| 183 |
+
"learning_rate": 5.1920529801324506e-05,
|
| 184 |
+
"loss": 1.1285,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.29,
|
| 189 |
+
"learning_rate": 5.059602649006623e-05,
|
| 190 |
+
"loss": 1.1556,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.34,
|
| 195 |
+
"learning_rate": 4.927152317880795e-05,
|
| 196 |
+
"loss": 1.0654,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.38,
|
| 201 |
+
"learning_rate": 4.794701986754967e-05,
|
| 202 |
+
"loss": 1.1231,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.43,
|
| 207 |
+
"learning_rate": 4.6622516556291395e-05,
|
| 208 |
+
"loss": 1.192,
|
| 209 |
+
"step": 320
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.47,
|
| 213 |
+
"learning_rate": 4.5298013245033114e-05,
|
| 214 |
+
"loss": 1.0347,
|
| 215 |
+
"step": 330
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.52,
|
| 219 |
+
"learning_rate": 4.397350993377484e-05,
|
| 220 |
+
"loss": 1.0308,
|
| 221 |
+
"step": 340
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.56,
|
| 225 |
+
"learning_rate": 4.264900662251656e-05,
|
| 226 |
+
"loss": 1.1021,
|
| 227 |
+
"step": 350
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.61,
|
| 231 |
+
"learning_rate": 4.132450331125828e-05,
|
| 232 |
+
"loss": 1.0571,
|
| 233 |
+
"step": 360
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.65,
|
| 237 |
+
"learning_rate": 4e-05,
|
| 238 |
+
"loss": 1.0717,
|
| 239 |
+
"step": 370
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.69,
|
| 243 |
+
"learning_rate": 3.867549668874173e-05,
|
| 244 |
+
"loss": 1.1333,
|
| 245 |
+
"step": 380
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.74,
|
| 249 |
+
"learning_rate": 3.735099337748345e-05,
|
| 250 |
+
"loss": 1.1149,
|
| 251 |
+
"step": 390
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.78,
|
| 255 |
+
"learning_rate": 3.6026490066225173e-05,
|
| 256 |
+
"loss": 1.0476,
|
| 257 |
+
"step": 400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.83,
|
| 261 |
+
"learning_rate": 3.470198675496689e-05,
|
| 262 |
+
"loss": 1.0465,
|
| 263 |
+
"step": 410
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.87,
|
| 267 |
+
"learning_rate": 3.337748344370861e-05,
|
| 268 |
+
"loss": 1.0967,
|
| 269 |
+
"step": 420
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.92,
|
| 273 |
+
"learning_rate": 3.205298013245033e-05,
|
| 274 |
+
"loss": 1.0408,
|
| 275 |
+
"step": 430
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.96,
|
| 279 |
+
"learning_rate": 3.0728476821192056e-05,
|
| 280 |
+
"loss": 1.03,
|
| 281 |
+
"step": 440
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 2.0,
|
| 285 |
+
"eval_accuracy": 0.6588186124268598,
|
| 286 |
+
"eval_loss": 0.9184884428977966,
|
| 287 |
+
"eval_runtime": 65.4725,
|
| 288 |
+
"eval_samples_per_second": 109.634,
|
| 289 |
+
"eval_steps_per_second": 3.437,
|
| 290 |
+
"step": 448
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.01,
|
| 294 |
+
"learning_rate": 2.9403973509933778e-05,
|
| 295 |
+
"loss": 1.0675,
|
| 296 |
+
"step": 450
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.05,
|
| 300 |
+
"learning_rate": 2.80794701986755e-05,
|
| 301 |
+
"loss": 1.0017,
|
| 302 |
+
"step": 460
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.1,
|
| 306 |
+
"learning_rate": 2.675496688741722e-05,
|
| 307 |
+
"loss": 1.0437,
|
| 308 |
+
"step": 470
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 2.14,
|
| 312 |
+
"learning_rate": 2.5430463576158945e-05,
|
| 313 |
+
"loss": 0.9995,
|
| 314 |
+
"step": 480
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.19,
|
| 318 |
+
"learning_rate": 2.4105960264900664e-05,
|
| 319 |
+
"loss": 1.0433,
|
| 320 |
+
"step": 490
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.23,
|
| 324 |
+
"learning_rate": 2.2781456953642386e-05,
|
| 325 |
+
"loss": 1.0132,
|
| 326 |
+
"step": 500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 2.28,
|
| 330 |
+
"learning_rate": 2.1456953642384105e-05,
|
| 331 |
+
"loss": 1.0186,
|
| 332 |
+
"step": 510
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.32,
|
| 336 |
+
"learning_rate": 2.013245033112583e-05,
|
| 337 |
+
"loss": 1.0136,
|
| 338 |
+
"step": 520
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.37,
|
| 342 |
+
"learning_rate": 1.8807947019867553e-05,
|
| 343 |
+
"loss": 1.0207,
|
| 344 |
+
"step": 530
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.41,
|
| 348 |
+
"learning_rate": 1.7483443708609275e-05,
|
| 349 |
+
"loss": 0.9965,
|
| 350 |
+
"step": 540
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.45,
|
| 354 |
+
"learning_rate": 1.6158940397350994e-05,
|
| 355 |
+
"loss": 0.9838,
|
| 356 |
+
"step": 550
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.5,
|
| 360 |
+
"learning_rate": 1.4834437086092716e-05,
|
| 361 |
+
"loss": 1.0405,
|
| 362 |
+
"step": 560
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.54,
|
| 366 |
+
"learning_rate": 1.3509933774834438e-05,
|
| 367 |
+
"loss": 0.9921,
|
| 368 |
+
"step": 570
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.59,
|
| 372 |
+
"learning_rate": 1.218543046357616e-05,
|
| 373 |
+
"loss": 1.0087,
|
| 374 |
+
"step": 580
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.63,
|
| 378 |
+
"learning_rate": 1.0860927152317881e-05,
|
| 379 |
+
"loss": 1.0011,
|
| 380 |
+
"step": 590
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.68,
|
| 384 |
+
"learning_rate": 9.536423841059603e-06,
|
| 385 |
+
"loss": 1.0205,
|
| 386 |
+
"step": 600
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.72,
|
| 390 |
+
"learning_rate": 8.211920529801326e-06,
|
| 391 |
+
"loss": 1.0155,
|
| 392 |
+
"step": 610
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.77,
|
| 396 |
+
"learning_rate": 6.887417218543046e-06,
|
| 397 |
+
"loss": 0.9777,
|
| 398 |
+
"step": 620
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.81,
|
| 402 |
+
"learning_rate": 5.5629139072847685e-06,
|
| 403 |
+
"loss": 0.9669,
|
| 404 |
+
"step": 630
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 2.86,
|
| 408 |
+
"learning_rate": 4.238410596026491e-06,
|
| 409 |
+
"loss": 0.9621,
|
| 410 |
+
"step": 640
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.9,
|
| 414 |
+
"learning_rate": 2.913907284768212e-06,
|
| 415 |
+
"loss": 0.9538,
|
| 416 |
+
"step": 650
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.94,
|
| 420 |
+
"learning_rate": 1.589403973509934e-06,
|
| 421 |
+
"loss": 1.0232,
|
| 422 |
+
"step": 660
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.99,
|
| 426 |
+
"learning_rate": 2.649006622516557e-07,
|
| 427 |
+
"loss": 0.9692,
|
| 428 |
+
"step": 670
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.0,
|
| 432 |
+
"eval_accuracy": 0.6851490665923655,
|
| 433 |
+
"eval_loss": 0.8512730002403259,
|
| 434 |
+
"eval_runtime": 67.5959,
|
| 435 |
+
"eval_samples_per_second": 106.19,
|
| 436 |
+
"eval_steps_per_second": 3.329,
|
| 437 |
+
"step": 672
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 3.0,
|
| 441 |
+
"step": 672,
|
| 442 |
+
"total_flos": 6.668732964123095e+18,
|
| 443 |
+
"train_loss": 1.145610089103381,
|
| 444 |
+
"train_runtime": 2105.9038,
|
| 445 |
+
"train_samples_per_second": 40.898,
|
| 446 |
+
"train_steps_per_second": 0.319
|
| 447 |
+
}
|
| 448 |
+
],
|
| 449 |
+
"max_steps": 672,
|
| 450 |
+
"num_train_epochs": 3,
|
| 451 |
+
"total_flos": 6.668732964123095e+18,
|
| 452 |
+
"trial_name": null,
|
| 453 |
+
"trial_params": null
|
| 454 |
+
}
|