End of training
Browse files- README.md +11 -1
- all_results.json +14 -8
- eval_results.json +8 -8
- runs/Nov07_17-36-37_gtown-28NZK54/events.out.tfevents.1731005439.gtown-28NZK54.23370.1 +3 -0
- train_results.json +6 -6
- trainer_state.json +822 -822
README.md
CHANGED
|
@@ -3,7 +3,11 @@ library_name: transformers
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: google/vit-base-patch16-224-in21k
|
| 5 |
tags:
|
|
|
|
|
|
|
| 6 |
- generated_from_trainer
|
|
|
|
|
|
|
| 7 |
model-index:
|
| 8 |
- name: vit-base-patch16-224-in21k
|
| 9 |
results: []
|
|
@@ -14,7 +18,13 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 14 |
|
| 15 |
# vit-base-patch16-224-in21k
|
| 16 |
|
| 17 |
-
This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
|
| 19 |
## Model description
|
| 20 |
|
|
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: google/vit-base-patch16-224-in21k
|
| 5 |
tags:
|
| 6 |
+
- image-classification
|
| 7 |
+
- vision
|
| 8 |
- generated_from_trainer
|
| 9 |
+
metrics:
|
| 10 |
+
- accuracy
|
| 11 |
model-index:
|
| 12 |
- name: vit-base-patch16-224-in21k
|
| 13 |
results: []
|
|
|
|
| 18 |
|
| 19 |
# vit-base-patch16-224-in21k
|
| 20 |
|
| 21 |
+
This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the chainyo/rvl-cdip dataset.
|
| 22 |
+
It achieves the following results on the evaluation set:
|
| 23 |
+
- Loss: 0.4223
|
| 24 |
+
- Accuracy: 0.8788
|
| 25 |
+
- Memory Allocated (gb): 1.49
|
| 26 |
+
- Max Memory Allocated (gb): 2.1
|
| 27 |
+
- Total Memory Available (gb): 126.62
|
| 28 |
|
| 29 |
## Model description
|
| 30 |
|
all_results.json
CHANGED
|
@@ -1,11 +1,17 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
|
|
|
| 4 |
"eval_model_preparation_time": 0.0078,
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 3.
|
| 8 |
-
"max_memory_allocated (GB)":
|
| 9 |
-
"memory_allocated (GB)":
|
| 10 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8787708333333333,
|
| 4 |
+
"eval_loss": 0.42233380675315857,
|
| 5 |
"eval_model_preparation_time": 0.0078,
|
| 6 |
+
"eval_runtime": 244.8398,
|
| 7 |
+
"eval_samples_per_second": 196.047,
|
| 8 |
+
"eval_steps_per_second": 3.063,
|
| 9 |
+
"max_memory_allocated (GB)": 2.1,
|
| 10 |
+
"memory_allocated (GB)": 1.49,
|
| 11 |
+
"total_flos": 6.324139790696448e+19,
|
| 12 |
+
"total_memory_available (GB)": 126.62,
|
| 13 |
+
"train_loss": 0.8088104343788297,
|
| 14 |
+
"train_runtime": 4169.6612,
|
| 15 |
+
"train_samples_per_second": 195.699,
|
| 16 |
+
"train_steps_per_second": 24.462
|
| 17 |
}
|
eval_results.json
CHANGED
|
@@ -1,11 +1,11 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"eval_runtime":
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 3.
|
| 8 |
-
"max_memory_allocated (GB)":
|
| 9 |
-
"memory_allocated (GB)":
|
| 10 |
"total_memory_available (GB)": 126.62
|
| 11 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8787708333333333,
|
| 4 |
+
"eval_loss": 0.42233380675315857,
|
| 5 |
+
"eval_runtime": 244.8398,
|
| 6 |
+
"eval_samples_per_second": 196.047,
|
| 7 |
+
"eval_steps_per_second": 3.063,
|
| 8 |
+
"max_memory_allocated (GB)": 2.1,
|
| 9 |
+
"memory_allocated (GB)": 1.49,
|
| 10 |
"total_memory_available (GB)": 126.62
|
| 11 |
}
|
runs/Nov07_17-36-37_gtown-28NZK54/events.out.tfevents.1731005439.gtown-28NZK54.23370.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:368115bda28a251ade5d2874097c5040c1cb5c5bebaa1b502a810676f9c52bcd
|
| 3 |
+
size 580
|
train_results.json
CHANGED
|
@@ -1,11 +1,11 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
-
"max_memory_allocated (GB)":
|
| 4 |
-
"memory_allocated (GB)":
|
| 5 |
"total_flos": 6.324139790696448e+19,
|
| 6 |
"total_memory_available (GB)": 126.62,
|
| 7 |
-
"train_loss": 0.
|
| 8 |
-
"train_runtime":
|
| 9 |
-
"train_samples_per_second":
|
| 10 |
-
"train_steps_per_second":
|
| 11 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
+
"max_memory_allocated (GB)": 2.1,
|
| 4 |
+
"memory_allocated (GB)": 1.46,
|
| 5 |
"total_flos": 6.324139790696448e+19,
|
| 6 |
"total_memory_available (GB)": 126.62,
|
| 7 |
+
"train_loss": 0.8088104343788297,
|
| 8 |
+
"train_runtime": 4169.6612,
|
| 9 |
+
"train_samples_per_second": 195.699,
|
| 10 |
+
"train_steps_per_second": 24.462
|
| 11 |
}
|
trainer_state.json
CHANGED
|
@@ -10,2055 +10,2055 @@
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.014705882352941176,
|
| 13 |
-
"grad_norm":
|
| 14 |
"learning_rate": 4.975490196078432e-05,
|
| 15 |
-
"loss":
|
| 16 |
-
"max_memory_allocated (GB)": 2.
|
| 17 |
-
"memory_allocated (GB)": 1.
|
| 18 |
"step": 500,
|
| 19 |
"total_memory_available (GB)": 126.62
|
| 20 |
},
|
| 21 |
{
|
| 22 |
"epoch": 0.029411764705882353,
|
| 23 |
-
"grad_norm":
|
| 24 |
"learning_rate": 4.9509803921568634e-05,
|
| 25 |
-
"loss":
|
| 26 |
-
"max_memory_allocated (GB)": 2.
|
| 27 |
-
"memory_allocated (GB)": 1.
|
| 28 |
"step": 1000,
|
| 29 |
"total_memory_available (GB)": 126.62
|
| 30 |
},
|
| 31 |
{
|
| 32 |
"epoch": 0.04411764705882353,
|
| 33 |
-
"grad_norm":
|
| 34 |
"learning_rate": 4.9264705882352944e-05,
|
| 35 |
-
"loss":
|
| 36 |
-
"max_memory_allocated (GB)": 2.
|
| 37 |
-
"memory_allocated (GB)": 1.
|
| 38 |
"step": 1500,
|
| 39 |
"total_memory_available (GB)": 126.62
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.058823529411764705,
|
| 43 |
-
"grad_norm":
|
| 44 |
"learning_rate": 4.901960784313725e-05,
|
| 45 |
-
"loss":
|
| 46 |
-
"max_memory_allocated (GB)": 2.
|
| 47 |
-
"memory_allocated (GB)": 1.
|
| 48 |
"step": 2000,
|
| 49 |
"total_memory_available (GB)": 126.62
|
| 50 |
},
|
| 51 |
{
|
| 52 |
"epoch": 0.07352941176470588,
|
| 53 |
-
"grad_norm":
|
| 54 |
"learning_rate": 4.877450980392157e-05,
|
| 55 |
-
"loss":
|
| 56 |
-
"max_memory_allocated (GB)": 2.
|
| 57 |
-
"memory_allocated (GB)": 1.
|
| 58 |
"step": 2500,
|
| 59 |
"total_memory_available (GB)": 126.62
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.08823529411764706,
|
| 63 |
-
"grad_norm":
|
| 64 |
"learning_rate": 4.8529411764705885e-05,
|
| 65 |
-
"loss":
|
| 66 |
-
"max_memory_allocated (GB)": 2.
|
| 67 |
-
"memory_allocated (GB)": 1.
|
| 68 |
"step": 3000,
|
| 69 |
"total_memory_available (GB)": 126.62
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.10294117647058823,
|
| 73 |
-
"grad_norm":
|
| 74 |
"learning_rate": 4.82843137254902e-05,
|
| 75 |
-
"loss":
|
| 76 |
-
"max_memory_allocated (GB)": 2.
|
| 77 |
-
"memory_allocated (GB)": 1.
|
| 78 |
"step": 3500,
|
| 79 |
"total_memory_available (GB)": 126.62
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.11764705882352941,
|
| 83 |
-
"grad_norm":
|
| 84 |
"learning_rate": 4.803921568627452e-05,
|
| 85 |
-
"loss":
|
| 86 |
-
"max_memory_allocated (GB)": 2.
|
| 87 |
-
"memory_allocated (GB)": 1.
|
| 88 |
"step": 4000,
|
| 89 |
"total_memory_available (GB)": 126.62
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 0.1323529411764706,
|
| 93 |
-
"grad_norm":
|
| 94 |
"learning_rate": 4.7794117647058826e-05,
|
| 95 |
-
"loss":
|
| 96 |
-
"max_memory_allocated (GB)": 2.
|
| 97 |
-
"memory_allocated (GB)": 1.
|
| 98 |
"step": 4500,
|
| 99 |
"total_memory_available (GB)": 126.62
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.14705882352941177,
|
| 103 |
-
"grad_norm":
|
| 104 |
"learning_rate": 4.7549019607843135e-05,
|
| 105 |
-
"loss":
|
| 106 |
-
"max_memory_allocated (GB)": 2.
|
| 107 |
-
"memory_allocated (GB)": 1.
|
| 108 |
"step": 5000,
|
| 109 |
"total_memory_available (GB)": 126.62
|
| 110 |
},
|
| 111 |
{
|
| 112 |
"epoch": 0.16176470588235295,
|
| 113 |
-
"grad_norm":
|
| 114 |
"learning_rate": 4.730392156862745e-05,
|
| 115 |
-
"loss":
|
| 116 |
-
"max_memory_allocated (GB)": 2.
|
| 117 |
-
"memory_allocated (GB)": 1.
|
| 118 |
"step": 5500,
|
| 119 |
"total_memory_available (GB)": 126.62
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 0.17647058823529413,
|
| 123 |
-
"grad_norm":
|
| 124 |
"learning_rate": 4.705882352941177e-05,
|
| 125 |
-
"loss":
|
| 126 |
-
"max_memory_allocated (GB)": 2.
|
| 127 |
-
"memory_allocated (GB)": 1.
|
| 128 |
"step": 6000,
|
| 129 |
"total_memory_available (GB)": 126.62
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.19117647058823528,
|
| 133 |
-
"grad_norm":
|
| 134 |
"learning_rate": 4.681372549019608e-05,
|
| 135 |
-
"loss":
|
| 136 |
-
"max_memory_allocated (GB)": 2.
|
| 137 |
-
"memory_allocated (GB)": 1.
|
| 138 |
"step": 6500,
|
| 139 |
"total_memory_available (GB)": 126.62
|
| 140 |
},
|
| 141 |
{
|
| 142 |
"epoch": 0.20588235294117646,
|
| 143 |
-
"grad_norm":
|
| 144 |
"learning_rate": 4.656862745098039e-05,
|
| 145 |
-
"loss":
|
| 146 |
-
"max_memory_allocated (GB)": 2.
|
| 147 |
-
"memory_allocated (GB)": 1.
|
| 148 |
"step": 7000,
|
| 149 |
"total_memory_available (GB)": 126.62
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.22058823529411764,
|
| 153 |
-
"grad_norm":
|
| 154 |
"learning_rate": 4.632352941176471e-05,
|
| 155 |
-
"loss":
|
| 156 |
-
"max_memory_allocated (GB)": 2.
|
| 157 |
-
"memory_allocated (GB)": 1.
|
| 158 |
"step": 7500,
|
| 159 |
"total_memory_available (GB)": 126.62
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.23529411764705882,
|
| 163 |
-
"grad_norm":
|
| 164 |
"learning_rate": 4.607843137254902e-05,
|
| 165 |
-
"loss":
|
| 166 |
-
"max_memory_allocated (GB)": 2.
|
| 167 |
-
"memory_allocated (GB)": 1.
|
| 168 |
"step": 8000,
|
| 169 |
"total_memory_available (GB)": 126.62
|
| 170 |
},
|
| 171 |
{
|
| 172 |
"epoch": 0.25,
|
| 173 |
-
"grad_norm":
|
| 174 |
"learning_rate": 4.5833333333333334e-05,
|
| 175 |
-
"loss":
|
| 176 |
-
"max_memory_allocated (GB)": 2.
|
| 177 |
-
"memory_allocated (GB)": 1.
|
| 178 |
"step": 8500,
|
| 179 |
"total_memory_available (GB)": 126.62
|
| 180 |
},
|
| 181 |
{
|
| 182 |
"epoch": 0.2647058823529412,
|
| 183 |
-
"grad_norm":
|
| 184 |
"learning_rate": 4.558823529411765e-05,
|
| 185 |
-
"loss":
|
| 186 |
-
"max_memory_allocated (GB)": 2.
|
| 187 |
-
"memory_allocated (GB)": 1.
|
| 188 |
"step": 9000,
|
| 189 |
"total_memory_available (GB)": 126.62
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.27941176470588236,
|
| 193 |
-
"grad_norm":
|
| 194 |
"learning_rate": 4.5343137254901966e-05,
|
| 195 |
-
"loss":
|
| 196 |
-
"max_memory_allocated (GB)": 2.
|
| 197 |
-
"memory_allocated (GB)": 1.
|
| 198 |
"step": 9500,
|
| 199 |
"total_memory_available (GB)": 126.62
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.29411764705882354,
|
| 203 |
-
"grad_norm":
|
| 204 |
"learning_rate": 4.5098039215686275e-05,
|
| 205 |
-
"loss":
|
| 206 |
-
"max_memory_allocated (GB)": 2.
|
| 207 |
-
"memory_allocated (GB)": 1.
|
| 208 |
"step": 10000,
|
| 209 |
"total_memory_available (GB)": 126.62
|
| 210 |
},
|
| 211 |
{
|
| 212 |
"epoch": 0.3088235294117647,
|
| 213 |
-
"grad_norm":
|
| 214 |
"learning_rate": 4.485294117647059e-05,
|
| 215 |
-
"loss":
|
| 216 |
-
"max_memory_allocated (GB)": 2.
|
| 217 |
-
"memory_allocated (GB)": 1.
|
| 218 |
"step": 10500,
|
| 219 |
"total_memory_available (GB)": 126.62
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.3235294117647059,
|
| 223 |
-
"grad_norm":
|
| 224 |
"learning_rate": 4.460784313725491e-05,
|
| 225 |
-
"loss":
|
| 226 |
-
"max_memory_allocated (GB)": 2.
|
| 227 |
-
"memory_allocated (GB)": 1.
|
| 228 |
"step": 11000,
|
| 229 |
"total_memory_available (GB)": 126.62
|
| 230 |
},
|
| 231 |
{
|
| 232 |
"epoch": 0.3382352941176471,
|
| 233 |
-
"grad_norm":
|
| 234 |
"learning_rate": 4.4362745098039216e-05,
|
| 235 |
-
"loss":
|
| 236 |
-
"max_memory_allocated (GB)": 2.
|
| 237 |
-
"memory_allocated (GB)": 1.
|
| 238 |
"step": 11500,
|
| 239 |
"total_memory_available (GB)": 126.62
|
| 240 |
},
|
| 241 |
{
|
| 242 |
"epoch": 0.35294117647058826,
|
| 243 |
-
"grad_norm":
|
| 244 |
"learning_rate": 4.411764705882353e-05,
|
| 245 |
-
"loss":
|
| 246 |
-
"max_memory_allocated (GB)": 2.
|
| 247 |
-
"memory_allocated (GB)": 1.
|
| 248 |
"step": 12000,
|
| 249 |
"total_memory_available (GB)": 126.62
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.36764705882352944,
|
| 253 |
-
"grad_norm":
|
| 254 |
"learning_rate": 4.387254901960784e-05,
|
| 255 |
-
"loss":
|
| 256 |
-
"max_memory_allocated (GB)": 2.
|
| 257 |
-
"memory_allocated (GB)": 1.
|
| 258 |
"step": 12500,
|
| 259 |
"total_memory_available (GB)": 126.62
|
| 260 |
},
|
| 261 |
{
|
| 262 |
"epoch": 0.38235294117647056,
|
| 263 |
-
"grad_norm":
|
| 264 |
"learning_rate": 4.362745098039216e-05,
|
| 265 |
-
"loss":
|
| 266 |
-
"max_memory_allocated (GB)": 2.
|
| 267 |
-
"memory_allocated (GB)": 1.
|
| 268 |
"step": 13000,
|
| 269 |
"total_memory_available (GB)": 126.62
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.39705882352941174,
|
| 273 |
-
"grad_norm":
|
| 274 |
"learning_rate": 4.3382352941176474e-05,
|
| 275 |
-
"loss":
|
| 276 |
-
"max_memory_allocated (GB)": 2.
|
| 277 |
-
"memory_allocated (GB)": 1.
|
| 278 |
"step": 13500,
|
| 279 |
"total_memory_available (GB)": 126.62
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.4117647058823529,
|
| 283 |
-
"grad_norm":
|
| 284 |
"learning_rate": 4.313725490196079e-05,
|
| 285 |
-
"loss":
|
| 286 |
-
"max_memory_allocated (GB)": 2.
|
| 287 |
-
"memory_allocated (GB)": 1.
|
| 288 |
"step": 14000,
|
| 289 |
"total_memory_available (GB)": 126.62
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"epoch": 0.4264705882352941,
|
| 293 |
-
"grad_norm":
|
| 294 |
"learning_rate": 4.28921568627451e-05,
|
| 295 |
-
"loss":
|
| 296 |
-
"max_memory_allocated (GB)": 2.
|
| 297 |
-
"memory_allocated (GB)": 1.
|
| 298 |
"step": 14500,
|
| 299 |
"total_memory_available (GB)": 126.62
|
| 300 |
},
|
| 301 |
{
|
| 302 |
"epoch": 0.4411764705882353,
|
| 303 |
-
"grad_norm":
|
| 304 |
"learning_rate": 4.2647058823529415e-05,
|
| 305 |
-
"loss":
|
| 306 |
-
"max_memory_allocated (GB)": 2.
|
| 307 |
-
"memory_allocated (GB)": 1.
|
| 308 |
"step": 15000,
|
| 309 |
"total_memory_available (GB)": 126.62
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.45588235294117646,
|
| 313 |
-
"grad_norm":
|
| 314 |
"learning_rate": 4.2401960784313724e-05,
|
| 315 |
-
"loss":
|
| 316 |
-
"max_memory_allocated (GB)": 2.
|
| 317 |
-
"memory_allocated (GB)": 1.
|
| 318 |
"step": 15500,
|
| 319 |
"total_memory_available (GB)": 126.62
|
| 320 |
},
|
| 321 |
{
|
| 322 |
"epoch": 0.47058823529411764,
|
| 323 |
-
"grad_norm":
|
| 324 |
"learning_rate": 4.215686274509804e-05,
|
| 325 |
-
"loss":
|
| 326 |
-
"max_memory_allocated (GB)": 2.
|
| 327 |
-
"memory_allocated (GB)": 1.
|
| 328 |
"step": 16000,
|
| 329 |
"total_memory_available (GB)": 126.62
|
| 330 |
},
|
| 331 |
{
|
| 332 |
"epoch": 0.4852941176470588,
|
| 333 |
-
"grad_norm":
|
| 334 |
"learning_rate": 4.1911764705882356e-05,
|
| 335 |
-
"loss":
|
| 336 |
-
"max_memory_allocated (GB)": 2.
|
| 337 |
-
"memory_allocated (GB)": 1.
|
| 338 |
"step": 16500,
|
| 339 |
"total_memory_available (GB)": 126.62
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.5,
|
| 343 |
-
"grad_norm":
|
| 344 |
"learning_rate": 4.166666666666667e-05,
|
| 345 |
-
"loss":
|
| 346 |
-
"max_memory_allocated (GB)": 2.
|
| 347 |
-
"memory_allocated (GB)": 1.
|
| 348 |
"step": 17000,
|
| 349 |
"total_memory_available (GB)": 126.62
|
| 350 |
},
|
| 351 |
{
|
| 352 |
"epoch": 0.5147058823529411,
|
| 353 |
-
"grad_norm":
|
| 354 |
"learning_rate": 4.142156862745099e-05,
|
| 355 |
-
"loss":
|
| 356 |
-
"max_memory_allocated (GB)": 2.
|
| 357 |
-
"memory_allocated (GB)": 1.
|
| 358 |
"step": 17500,
|
| 359 |
"total_memory_available (GB)": 126.62
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"epoch": 0.5294117647058824,
|
| 363 |
-
"grad_norm":
|
| 364 |
"learning_rate": 4.11764705882353e-05,
|
| 365 |
-
"loss":
|
| 366 |
-
"max_memory_allocated (GB)": 2.
|
| 367 |
-
"memory_allocated (GB)": 1.
|
| 368 |
"step": 18000,
|
| 369 |
"total_memory_available (GB)": 126.62
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.5441176470588235,
|
| 373 |
-
"grad_norm":
|
| 374 |
"learning_rate": 4.0931372549019607e-05,
|
| 375 |
-
"loss":
|
| 376 |
-
"max_memory_allocated (GB)": 2.
|
| 377 |
-
"memory_allocated (GB)": 1.
|
| 378 |
"step": 18500,
|
| 379 |
"total_memory_available (GB)": 126.62
|
| 380 |
},
|
| 381 |
{
|
| 382 |
"epoch": 0.5588235294117647,
|
| 383 |
-
"grad_norm":
|
| 384 |
"learning_rate": 4.068627450980392e-05,
|
| 385 |
-
"loss":
|
| 386 |
-
"max_memory_allocated (GB)": 2.
|
| 387 |
-
"memory_allocated (GB)": 1.
|
| 388 |
"step": 19000,
|
| 389 |
"total_memory_available (GB)": 126.62
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 0.5735294117647058,
|
| 393 |
-
"grad_norm":
|
| 394 |
"learning_rate": 4.044117647058824e-05,
|
| 395 |
-
"loss":
|
| 396 |
-
"max_memory_allocated (GB)": 2.
|
| 397 |
-
"memory_allocated (GB)": 1.
|
| 398 |
"step": 19500,
|
| 399 |
"total_memory_available (GB)": 126.62
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.5882352941176471,
|
| 403 |
-
"grad_norm":
|
| 404 |
"learning_rate": 4.0196078431372555e-05,
|
| 405 |
-
"loss":
|
| 406 |
-
"max_memory_allocated (GB)": 2.
|
| 407 |
-
"memory_allocated (GB)": 1.
|
| 408 |
"step": 20000,
|
| 409 |
"total_memory_available (GB)": 126.62
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 0.6029411764705882,
|
| 413 |
-
"grad_norm":
|
| 414 |
"learning_rate": 3.9950980392156864e-05,
|
| 415 |
-
"loss":
|
| 416 |
-
"max_memory_allocated (GB)": 2.
|
| 417 |
-
"memory_allocated (GB)": 1.
|
| 418 |
"step": 20500,
|
| 419 |
"total_memory_available (GB)": 126.62
|
| 420 |
},
|
| 421 |
{
|
| 422 |
"epoch": 0.6176470588235294,
|
| 423 |
-
"grad_norm":
|
| 424 |
"learning_rate": 3.970588235294117e-05,
|
| 425 |
-
"loss":
|
| 426 |
-
"max_memory_allocated (GB)": 2.
|
| 427 |
-
"memory_allocated (GB)": 1.
|
| 428 |
"step": 21000,
|
| 429 |
"total_memory_available (GB)": 126.62
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.6323529411764706,
|
| 433 |
-
"grad_norm":
|
| 434 |
"learning_rate": 3.946078431372549e-05,
|
| 435 |
-
"loss":
|
| 436 |
-
"max_memory_allocated (GB)": 2.
|
| 437 |
-
"memory_allocated (GB)": 1.
|
| 438 |
"step": 21500,
|
| 439 |
"total_memory_available (GB)": 126.62
|
| 440 |
},
|
| 441 |
{
|
| 442 |
"epoch": 0.6470588235294118,
|
| 443 |
-
"grad_norm":
|
| 444 |
"learning_rate": 3.9215686274509805e-05,
|
| 445 |
-
"loss":
|
| 446 |
-
"max_memory_allocated (GB)": 2.
|
| 447 |
-
"memory_allocated (GB)": 1.
|
| 448 |
"step": 22000,
|
| 449 |
"total_memory_available (GB)": 126.62
|
| 450 |
},
|
| 451 |
{
|
| 452 |
"epoch": 0.6617647058823529,
|
| 453 |
-
"grad_norm":
|
| 454 |
"learning_rate": 3.897058823529412e-05,
|
| 455 |
-
"loss":
|
| 456 |
-
"max_memory_allocated (GB)": 2.
|
| 457 |
-
"memory_allocated (GB)": 1.
|
| 458 |
"step": 22500,
|
| 459 |
"total_memory_available (GB)": 126.62
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.6764705882352942,
|
| 463 |
-
"grad_norm":
|
| 464 |
"learning_rate": 3.872549019607844e-05,
|
| 465 |
-
"loss":
|
| 466 |
-
"max_memory_allocated (GB)": 2.
|
| 467 |
-
"memory_allocated (GB)": 1.
|
| 468 |
"step": 23000,
|
| 469 |
"total_memory_available (GB)": 126.62
|
| 470 |
},
|
| 471 |
{
|
| 472 |
"epoch": 0.6911764705882353,
|
| 473 |
-
"grad_norm":
|
| 474 |
"learning_rate": 3.8480392156862746e-05,
|
| 475 |
-
"loss":
|
| 476 |
-
"max_memory_allocated (GB)": 2.
|
| 477 |
-
"memory_allocated (GB)": 1.
|
| 478 |
"step": 23500,
|
| 479 |
"total_memory_available (GB)": 126.62
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 0.7058823529411765,
|
| 483 |
-
"grad_norm":
|
| 484 |
"learning_rate": 3.8235294117647055e-05,
|
| 485 |
-
"loss":
|
| 486 |
-
"max_memory_allocated (GB)": 2.
|
| 487 |
-
"memory_allocated (GB)": 1.
|
| 488 |
"step": 24000,
|
| 489 |
"total_memory_available (GB)": 126.62
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.7205882352941176,
|
| 493 |
-
"grad_norm":
|
| 494 |
"learning_rate": 3.799019607843137e-05,
|
| 495 |
-
"loss":
|
| 496 |
-
"max_memory_allocated (GB)": 2.
|
| 497 |
-
"memory_allocated (GB)": 1.
|
| 498 |
"step": 24500,
|
| 499 |
"total_memory_available (GB)": 126.62
|
| 500 |
},
|
| 501 |
{
|
| 502 |
"epoch": 0.7352941176470589,
|
| 503 |
-
"grad_norm":
|
| 504 |
"learning_rate": 3.774509803921569e-05,
|
| 505 |
-
"loss":
|
| 506 |
-
"max_memory_allocated (GB)": 2.
|
| 507 |
-
"memory_allocated (GB)": 1.
|
| 508 |
"step": 25000,
|
| 509 |
"total_memory_available (GB)": 126.62
|
| 510 |
},
|
| 511 |
{
|
| 512 |
"epoch": 0.75,
|
| 513 |
-
"grad_norm":
|
| 514 |
"learning_rate": 3.7500000000000003e-05,
|
| 515 |
-
"loss":
|
| 516 |
-
"max_memory_allocated (GB)": 2.
|
| 517 |
-
"memory_allocated (GB)": 1.
|
| 518 |
"step": 25500,
|
| 519 |
"total_memory_available (GB)": 126.62
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.7647058823529411,
|
| 523 |
-
"grad_norm":
|
| 524 |
"learning_rate": 3.725490196078432e-05,
|
| 525 |
-
"loss":
|
| 526 |
-
"max_memory_allocated (GB)": 2.
|
| 527 |
-
"memory_allocated (GB)": 1.
|
| 528 |
"step": 26000,
|
| 529 |
"total_memory_available (GB)": 126.62
|
| 530 |
},
|
| 531 |
{
|
| 532 |
"epoch": 0.7794117647058824,
|
| 533 |
-
"grad_norm":
|
| 534 |
"learning_rate": 3.700980392156863e-05,
|
| 535 |
-
"loss":
|
| 536 |
-
"max_memory_allocated (GB)": 2.
|
| 537 |
-
"memory_allocated (GB)": 1.
|
| 538 |
"step": 26500,
|
| 539 |
"total_memory_available (GB)": 126.62
|
| 540 |
},
|
| 541 |
{
|
| 542 |
"epoch": 0.7941176470588235,
|
| 543 |
-
"grad_norm":
|
| 544 |
"learning_rate": 3.6764705882352945e-05,
|
| 545 |
-
"loss":
|
| 546 |
-
"max_memory_allocated (GB)": 2.
|
| 547 |
-
"memory_allocated (GB)": 1.
|
| 548 |
"step": 27000,
|
| 549 |
"total_memory_available (GB)": 126.62
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.8088235294117647,
|
| 553 |
-
"grad_norm":
|
| 554 |
"learning_rate": 3.6519607843137254e-05,
|
| 555 |
-
"loss":
|
| 556 |
-
"max_memory_allocated (GB)": 2.
|
| 557 |
-
"memory_allocated (GB)": 1.
|
| 558 |
"step": 27500,
|
| 559 |
"total_memory_available (GB)": 126.62
|
| 560 |
},
|
| 561 |
{
|
| 562 |
"epoch": 0.8235294117647058,
|
| 563 |
-
"grad_norm":
|
| 564 |
"learning_rate": 3.627450980392157e-05,
|
| 565 |
-
"loss":
|
| 566 |
-
"max_memory_allocated (GB)": 2.
|
| 567 |
-
"memory_allocated (GB)": 1.
|
| 568 |
"step": 28000,
|
| 569 |
"total_memory_available (GB)": 126.62
|
| 570 |
},
|
| 571 |
{
|
| 572 |
"epoch": 0.8382352941176471,
|
| 573 |
-
"grad_norm":
|
| 574 |
"learning_rate": 3.6029411764705886e-05,
|
| 575 |
-
"loss":
|
| 576 |
-
"max_memory_allocated (GB)": 2.
|
| 577 |
-
"memory_allocated (GB)": 1.
|
| 578 |
"step": 28500,
|
| 579 |
"total_memory_available (GB)": 126.62
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.8529411764705882,
|
| 583 |
-
"grad_norm":
|
| 584 |
"learning_rate": 3.5784313725490195e-05,
|
| 585 |
-
"loss":
|
| 586 |
-
"max_memory_allocated (GB)": 2.
|
| 587 |
-
"memory_allocated (GB)": 1.
|
| 588 |
"step": 29000,
|
| 589 |
"total_memory_available (GB)": 126.62
|
| 590 |
},
|
| 591 |
{
|
| 592 |
"epoch": 0.8676470588235294,
|
| 593 |
-
"grad_norm":
|
| 594 |
"learning_rate": 3.553921568627451e-05,
|
| 595 |
-
"loss":
|
| 596 |
-
"max_memory_allocated (GB)": 2.
|
| 597 |
-
"memory_allocated (GB)": 1.
|
| 598 |
"step": 29500,
|
| 599 |
"total_memory_available (GB)": 126.62
|
| 600 |
},
|
| 601 |
{
|
| 602 |
"epoch": 0.8823529411764706,
|
| 603 |
-
"grad_norm":
|
| 604 |
"learning_rate": 3.529411764705883e-05,
|
| 605 |
-
"loss":
|
| 606 |
-
"max_memory_allocated (GB)": 2.
|
| 607 |
-
"memory_allocated (GB)": 1.
|
| 608 |
"step": 30000,
|
| 609 |
"total_memory_available (GB)": 126.62
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.8970588235294118,
|
| 613 |
-
"grad_norm":
|
| 614 |
"learning_rate": 3.5049019607843136e-05,
|
| 615 |
-
"loss":
|
| 616 |
-
"max_memory_allocated (GB)": 2.
|
| 617 |
-
"memory_allocated (GB)": 1.
|
| 618 |
"step": 30500,
|
| 619 |
"total_memory_available (GB)": 126.62
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 0.9117647058823529,
|
| 623 |
-
"grad_norm":
|
| 624 |
"learning_rate": 3.480392156862745e-05,
|
| 625 |
-
"loss":
|
| 626 |
-
"max_memory_allocated (GB)": 2.
|
| 627 |
-
"memory_allocated (GB)": 1.
|
| 628 |
"step": 31000,
|
| 629 |
"total_memory_available (GB)": 126.62
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 0.9264705882352942,
|
| 633 |
-
"grad_norm":
|
| 634 |
"learning_rate": 3.455882352941177e-05,
|
| 635 |
-
"loss":
|
| 636 |
-
"max_memory_allocated (GB)": 2.
|
| 637 |
-
"memory_allocated (GB)": 1.
|
| 638 |
"step": 31500,
|
| 639 |
"total_memory_available (GB)": 126.62
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.9411764705882353,
|
| 643 |
-
"grad_norm":
|
| 644 |
"learning_rate": 3.431372549019608e-05,
|
| 645 |
-
"loss":
|
| 646 |
-
"max_memory_allocated (GB)": 2.
|
| 647 |
-
"memory_allocated (GB)": 1.
|
| 648 |
"step": 32000,
|
| 649 |
"total_memory_available (GB)": 126.62
|
| 650 |
},
|
| 651 |
{
|
| 652 |
"epoch": 0.9558823529411765,
|
| 653 |
-
"grad_norm":
|
| 654 |
"learning_rate": 3.4068627450980394e-05,
|
| 655 |
-
"loss":
|
| 656 |
-
"max_memory_allocated (GB)": 2.
|
| 657 |
-
"memory_allocated (GB)": 1.
|
| 658 |
"step": 32500,
|
| 659 |
"total_memory_available (GB)": 126.62
|
| 660 |
},
|
| 661 |
{
|
| 662 |
"epoch": 0.9705882352941176,
|
| 663 |
-
"grad_norm":
|
| 664 |
"learning_rate": 3.382352941176471e-05,
|
| 665 |
-
"loss":
|
| 666 |
-
"max_memory_allocated (GB)": 2.
|
| 667 |
-
"memory_allocated (GB)": 1.
|
| 668 |
"step": 33000,
|
| 669 |
"total_memory_available (GB)": 126.62
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.9852941176470589,
|
| 673 |
-
"grad_norm":
|
| 674 |
"learning_rate": 3.357843137254902e-05,
|
| 675 |
-
"loss":
|
| 676 |
-
"max_memory_allocated (GB)": 2.
|
| 677 |
-
"memory_allocated (GB)": 1.
|
| 678 |
"step": 33500,
|
| 679 |
"total_memory_available (GB)": 126.62
|
| 680 |
},
|
| 681 |
{
|
| 682 |
"epoch": 1.0,
|
| 683 |
-
"grad_norm":
|
| 684 |
"learning_rate": 3.3333333333333335e-05,
|
| 685 |
-
"loss":
|
| 686 |
-
"max_memory_allocated (GB)": 2.
|
| 687 |
-
"memory_allocated (GB)": 1.
|
| 688 |
"step": 34000,
|
| 689 |
"total_memory_available (GB)": 126.62
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 1.0147058823529411,
|
| 693 |
-
"grad_norm":
|
| 694 |
"learning_rate": 3.308823529411765e-05,
|
| 695 |
-
"loss":
|
| 696 |
-
"max_memory_allocated (GB)": 2.
|
| 697 |
-
"memory_allocated (GB)": 1.
|
| 698 |
"step": 34500,
|
| 699 |
"total_memory_available (GB)": 126.62
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 1.0294117647058822,
|
| 703 |
-
"grad_norm":
|
| 704 |
"learning_rate": 3.284313725490196e-05,
|
| 705 |
-
"loss":
|
| 706 |
-
"max_memory_allocated (GB)": 2.
|
| 707 |
-
"memory_allocated (GB)": 1.
|
| 708 |
"step": 35000,
|
| 709 |
"total_memory_available (GB)": 126.62
|
| 710 |
},
|
| 711 |
{
|
| 712 |
"epoch": 1.0441176470588236,
|
| 713 |
-
"grad_norm":
|
| 714 |
"learning_rate": 3.2598039215686276e-05,
|
| 715 |
-
"loss":
|
| 716 |
-
"max_memory_allocated (GB)": 2.
|
| 717 |
-
"memory_allocated (GB)": 1.
|
| 718 |
"step": 35500,
|
| 719 |
"total_memory_available (GB)": 126.62
|
| 720 |
},
|
| 721 |
{
|
| 722 |
"epoch": 1.0588235294117647,
|
| 723 |
-
"grad_norm":
|
| 724 |
"learning_rate": 3.235294117647059e-05,
|
| 725 |
-
"loss":
|
| 726 |
-
"max_memory_allocated (GB)": 2.
|
| 727 |
-
"memory_allocated (GB)": 1.
|
| 728 |
"step": 36000,
|
| 729 |
"total_memory_available (GB)": 126.62
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 1.0735294117647058,
|
| 733 |
-
"grad_norm":
|
| 734 |
"learning_rate": 3.210784313725491e-05,
|
| 735 |
-
"loss":
|
| 736 |
-
"max_memory_allocated (GB)": 2.
|
| 737 |
-
"memory_allocated (GB)": 1.
|
| 738 |
"step": 36500,
|
| 739 |
"total_memory_available (GB)": 126.62
|
| 740 |
},
|
| 741 |
{
|
| 742 |
"epoch": 1.088235294117647,
|
| 743 |
-
"grad_norm":
|
| 744 |
"learning_rate": 3.186274509803922e-05,
|
| 745 |
-
"loss":
|
| 746 |
-
"max_memory_allocated (GB)": 2.
|
| 747 |
-
"memory_allocated (GB)": 1.
|
| 748 |
"step": 37000,
|
| 749 |
"total_memory_available (GB)": 126.62
|
| 750 |
},
|
| 751 |
{
|
| 752 |
"epoch": 1.1029411764705883,
|
| 753 |
-
"grad_norm":
|
| 754 |
"learning_rate": 3.161764705882353e-05,
|
| 755 |
-
"loss":
|
| 756 |
-
"max_memory_allocated (GB)": 2.
|
| 757 |
-
"memory_allocated (GB)": 1.
|
| 758 |
"step": 37500,
|
| 759 |
"total_memory_available (GB)": 126.62
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 1.1176470588235294,
|
| 763 |
-
"grad_norm":
|
| 764 |
"learning_rate": 3.137254901960784e-05,
|
| 765 |
-
"loss":
|
| 766 |
-
"max_memory_allocated (GB)": 2.
|
| 767 |
-
"memory_allocated (GB)": 1.
|
| 768 |
"step": 38000,
|
| 769 |
"total_memory_available (GB)": 126.62
|
| 770 |
},
|
| 771 |
{
|
| 772 |
"epoch": 1.1323529411764706,
|
| 773 |
-
"grad_norm":
|
| 774 |
"learning_rate": 3.112745098039216e-05,
|
| 775 |
-
"loss":
|
| 776 |
-
"max_memory_allocated (GB)": 2.
|
| 777 |
-
"memory_allocated (GB)": 1.
|
| 778 |
"step": 38500,
|
| 779 |
"total_memory_available (GB)": 126.62
|
| 780 |
},
|
| 781 |
{
|
| 782 |
"epoch": 1.1470588235294117,
|
| 783 |
-
"grad_norm":
|
| 784 |
"learning_rate": 3.0882352941176475e-05,
|
| 785 |
-
"loss":
|
| 786 |
-
"max_memory_allocated (GB)": 2.
|
| 787 |
-
"memory_allocated (GB)": 1.
|
| 788 |
"step": 39000,
|
| 789 |
"total_memory_available (GB)": 126.62
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 1.161764705882353,
|
| 793 |
-
"grad_norm":
|
| 794 |
"learning_rate": 3.063725490196079e-05,
|
| 795 |
-
"loss":
|
| 796 |
-
"max_memory_allocated (GB)": 2.
|
| 797 |
-
"memory_allocated (GB)": 1.
|
| 798 |
"step": 39500,
|
| 799 |
"total_memory_available (GB)": 126.62
|
| 800 |
},
|
| 801 |
{
|
| 802 |
"epoch": 1.1764705882352942,
|
| 803 |
-
"grad_norm":
|
| 804 |
"learning_rate": 3.0392156862745097e-05,
|
| 805 |
-
"loss":
|
| 806 |
-
"max_memory_allocated (GB)": 2.
|
| 807 |
-
"memory_allocated (GB)": 1.
|
| 808 |
"step": 40000,
|
| 809 |
"total_memory_available (GB)": 126.62
|
| 810 |
},
|
| 811 |
{
|
| 812 |
"epoch": 1.1911764705882353,
|
| 813 |
-
"grad_norm":
|
| 814 |
"learning_rate": 3.0147058823529413e-05,
|
| 815 |
-
"loss":
|
| 816 |
-
"max_memory_allocated (GB)": 2.
|
| 817 |
-
"memory_allocated (GB)": 1.
|
| 818 |
"step": 40500,
|
| 819 |
"total_memory_available (GB)": 126.62
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 1.2058823529411764,
|
| 823 |
-
"grad_norm":
|
| 824 |
"learning_rate": 2.9901960784313725e-05,
|
| 825 |
-
"loss":
|
| 826 |
-
"max_memory_allocated (GB)": 2.
|
| 827 |
-
"memory_allocated (GB)": 1.
|
| 828 |
"step": 41000,
|
| 829 |
"total_memory_available (GB)": 126.62
|
| 830 |
},
|
| 831 |
{
|
| 832 |
"epoch": 1.2205882352941178,
|
| 833 |
-
"grad_norm":
|
| 834 |
"learning_rate": 2.965686274509804e-05,
|
| 835 |
-
"loss":
|
| 836 |
-
"max_memory_allocated (GB)": 2.
|
| 837 |
-
"memory_allocated (GB)": 1.
|
| 838 |
"step": 41500,
|
| 839 |
"total_memory_available (GB)": 126.62
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 1.2352941176470589,
|
| 843 |
-
"grad_norm":
|
| 844 |
"learning_rate": 2.9411764705882354e-05,
|
| 845 |
-
"loss":
|
| 846 |
-
"max_memory_allocated (GB)": 2.
|
| 847 |
-
"memory_allocated (GB)": 1.
|
| 848 |
"step": 42000,
|
| 849 |
"total_memory_available (GB)": 126.62
|
| 850 |
},
|
| 851 |
{
|
| 852 |
"epoch": 1.25,
|
| 853 |
-
"grad_norm":
|
| 854 |
"learning_rate": 2.916666666666667e-05,
|
| 855 |
-
"loss":
|
| 856 |
-
"max_memory_allocated (GB)": 2.
|
| 857 |
-
"memory_allocated (GB)": 1.
|
| 858 |
"step": 42500,
|
| 859 |
"total_memory_available (GB)": 126.62
|
| 860 |
},
|
| 861 |
{
|
| 862 |
"epoch": 1.2647058823529411,
|
| 863 |
-
"grad_norm":
|
| 864 |
"learning_rate": 2.8921568627450986e-05,
|
| 865 |
-
"loss":
|
| 866 |
-
"max_memory_allocated (GB)": 2.
|
| 867 |
-
"memory_allocated (GB)": 1.
|
| 868 |
"step": 43000,
|
| 869 |
"total_memory_available (GB)": 126.62
|
| 870 |
},
|
| 871 |
{
|
| 872 |
"epoch": 1.2794117647058822,
|
| 873 |
-
"grad_norm":
|
| 874 |
"learning_rate": 2.8676470588235295e-05,
|
| 875 |
-
"loss":
|
| 876 |
-
"max_memory_allocated (GB)": 2.
|
| 877 |
-
"memory_allocated (GB)": 1.
|
| 878 |
"step": 43500,
|
| 879 |
"total_memory_available (GB)": 126.62
|
| 880 |
},
|
| 881 |
{
|
| 882 |
"epoch": 1.2941176470588236,
|
| 883 |
-
"grad_norm":
|
| 884 |
"learning_rate": 2.8431372549019608e-05,
|
| 885 |
-
"loss":
|
| 886 |
-
"max_memory_allocated (GB)": 2.
|
| 887 |
-
"memory_allocated (GB)": 1.
|
| 888 |
"step": 44000,
|
| 889 |
"total_memory_available (GB)": 126.62
|
| 890 |
},
|
| 891 |
{
|
| 892 |
"epoch": 1.3088235294117647,
|
| 893 |
-
"grad_norm":
|
| 894 |
"learning_rate": 2.8186274509803924e-05,
|
| 895 |
-
"loss":
|
| 896 |
-
"max_memory_allocated (GB)": 2.
|
| 897 |
-
"memory_allocated (GB)": 1.
|
| 898 |
"step": 44500,
|
| 899 |
"total_memory_available (GB)": 126.62
|
| 900 |
},
|
| 901 |
{
|
| 902 |
"epoch": 1.3235294117647058,
|
| 903 |
-
"grad_norm":
|
| 904 |
"learning_rate": 2.7941176470588236e-05,
|
| 905 |
-
"loss":
|
| 906 |
-
"max_memory_allocated (GB)": 2.
|
| 907 |
-
"memory_allocated (GB)": 1.
|
| 908 |
"step": 45000,
|
| 909 |
"total_memory_available (GB)": 126.62
|
| 910 |
},
|
| 911 |
{
|
| 912 |
"epoch": 1.3382352941176472,
|
| 913 |
-
"grad_norm":
|
| 914 |
"learning_rate": 2.7696078431372552e-05,
|
| 915 |
-
"loss":
|
| 916 |
-
"max_memory_allocated (GB)": 2.
|
| 917 |
-
"memory_allocated (GB)": 1.
|
| 918 |
"step": 45500,
|
| 919 |
"total_memory_available (GB)": 126.62
|
| 920 |
},
|
| 921 |
{
|
| 922 |
"epoch": 1.3529411764705883,
|
| 923 |
-
"grad_norm":
|
| 924 |
"learning_rate": 2.7450980392156865e-05,
|
| 925 |
-
"loss":
|
| 926 |
-
"max_memory_allocated (GB)": 2.
|
| 927 |
-
"memory_allocated (GB)": 1.
|
| 928 |
"step": 46000,
|
| 929 |
"total_memory_available (GB)": 126.62
|
| 930 |
},
|
| 931 |
{
|
| 932 |
"epoch": 1.3676470588235294,
|
| 933 |
-
"grad_norm":
|
| 934 |
"learning_rate": 2.7205882352941174e-05,
|
| 935 |
-
"loss":
|
| 936 |
-
"max_memory_allocated (GB)": 2.
|
| 937 |
-
"memory_allocated (GB)": 1.
|
| 938 |
"step": 46500,
|
| 939 |
"total_memory_available (GB)": 126.62
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 1.3823529411764706,
|
| 943 |
-
"grad_norm":
|
| 944 |
"learning_rate": 2.696078431372549e-05,
|
| 945 |
-
"loss":
|
| 946 |
-
"max_memory_allocated (GB)": 2.
|
| 947 |
-
"memory_allocated (GB)": 1.
|
| 948 |
"step": 47000,
|
| 949 |
"total_memory_available (GB)": 126.62
|
| 950 |
},
|
| 951 |
{
|
| 952 |
"epoch": 1.3970588235294117,
|
| 953 |
-
"grad_norm":
|
| 954 |
"learning_rate": 2.6715686274509806e-05,
|
| 955 |
-
"loss":
|
| 956 |
-
"max_memory_allocated (GB)": 2.
|
| 957 |
-
"memory_allocated (GB)": 1.
|
| 958 |
"step": 47500,
|
| 959 |
"total_memory_available (GB)": 126.62
|
| 960 |
},
|
| 961 |
{
|
| 962 |
"epoch": 1.4117647058823528,
|
| 963 |
-
"grad_norm":
|
| 964 |
"learning_rate": 2.647058823529412e-05,
|
| 965 |
-
"loss":
|
| 966 |
-
"max_memory_allocated (GB)": 2.
|
| 967 |
-
"memory_allocated (GB)": 1.
|
| 968 |
"step": 48000,
|
| 969 |
"total_memory_available (GB)": 126.62
|
| 970 |
},
|
| 971 |
{
|
| 972 |
"epoch": 1.4264705882352942,
|
| 973 |
-
"grad_norm":
|
| 974 |
"learning_rate": 2.6225490196078435e-05,
|
| 975 |
-
"loss":
|
| 976 |
-
"max_memory_allocated (GB)": 2.
|
| 977 |
-
"memory_allocated (GB)": 1.
|
| 978 |
"step": 48500,
|
| 979 |
"total_memory_available (GB)": 126.62
|
| 980 |
},
|
| 981 |
{
|
| 982 |
"epoch": 1.4411764705882353,
|
| 983 |
-
"grad_norm":
|
| 984 |
"learning_rate": 2.5980392156862747e-05,
|
| 985 |
-
"loss":
|
| 986 |
-
"max_memory_allocated (GB)": 2.
|
| 987 |
-
"memory_allocated (GB)": 1.
|
| 988 |
"step": 49000,
|
| 989 |
"total_memory_available (GB)": 126.62
|
| 990 |
},
|
| 991 |
{
|
| 992 |
"epoch": 1.4558823529411764,
|
| 993 |
-
"grad_norm":
|
| 994 |
"learning_rate": 2.5735294117647057e-05,
|
| 995 |
-
"loss":
|
| 996 |
-
"max_memory_allocated (GB)": 2.
|
| 997 |
-
"memory_allocated (GB)": 1.
|
| 998 |
"step": 49500,
|
| 999 |
"total_memory_available (GB)": 126.62
|
| 1000 |
},
|
| 1001 |
{
|
| 1002 |
"epoch": 1.4705882352941178,
|
| 1003 |
-
"grad_norm":
|
| 1004 |
"learning_rate": 2.5490196078431373e-05,
|
| 1005 |
-
"loss":
|
| 1006 |
-
"max_memory_allocated (GB)": 2.
|
| 1007 |
-
"memory_allocated (GB)": 1.
|
| 1008 |
"step": 50000,
|
| 1009 |
"total_memory_available (GB)": 126.62
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
"epoch": 1.4852941176470589,
|
| 1013 |
-
"grad_norm":
|
| 1014 |
"learning_rate": 2.5245098039215685e-05,
|
| 1015 |
-
"loss":
|
| 1016 |
-
"max_memory_allocated (GB)": 2.
|
| 1017 |
-
"memory_allocated (GB)": 1.
|
| 1018 |
"step": 50500,
|
| 1019 |
"total_memory_available (GB)": 126.62
|
| 1020 |
},
|
| 1021 |
{
|
| 1022 |
"epoch": 1.5,
|
| 1023 |
-
"grad_norm":
|
| 1024 |
"learning_rate": 2.5e-05,
|
| 1025 |
-
"loss":
|
| 1026 |
-
"max_memory_allocated (GB)": 2.
|
| 1027 |
-
"memory_allocated (GB)": 1.
|
| 1028 |
"step": 51000,
|
| 1029 |
"total_memory_available (GB)": 126.62
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
"epoch": 1.5147058823529411,
|
| 1033 |
-
"grad_norm":
|
| 1034 |
"learning_rate": 2.4754901960784317e-05,
|
| 1035 |
-
"loss":
|
| 1036 |
-
"max_memory_allocated (GB)": 2.
|
| 1037 |
-
"memory_allocated (GB)": 1.
|
| 1038 |
"step": 51500,
|
| 1039 |
"total_memory_available (GB)": 126.62
|
| 1040 |
},
|
| 1041 |
{
|
| 1042 |
"epoch": 1.5294117647058822,
|
| 1043 |
-
"grad_norm":
|
| 1044 |
"learning_rate": 2.4509803921568626e-05,
|
| 1045 |
-
"loss":
|
| 1046 |
-
"max_memory_allocated (GB)": 2.
|
| 1047 |
-
"memory_allocated (GB)": 1.
|
| 1048 |
"step": 52000,
|
| 1049 |
"total_memory_available (GB)": 126.62
|
| 1050 |
},
|
| 1051 |
{
|
| 1052 |
"epoch": 1.5441176470588234,
|
| 1053 |
-
"grad_norm":
|
| 1054 |
"learning_rate": 2.4264705882352942e-05,
|
| 1055 |
-
"loss":
|
| 1056 |
-
"max_memory_allocated (GB)": 2.
|
| 1057 |
-
"memory_allocated (GB)": 1.
|
| 1058 |
"step": 52500,
|
| 1059 |
"total_memory_available (GB)": 126.62
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
"epoch": 1.5588235294117647,
|
| 1063 |
-
"grad_norm":
|
| 1064 |
"learning_rate": 2.401960784313726e-05,
|
| 1065 |
-
"loss":
|
| 1066 |
-
"max_memory_allocated (GB)": 2.
|
| 1067 |
-
"memory_allocated (GB)": 1.
|
| 1068 |
"step": 53000,
|
| 1069 |
"total_memory_available (GB)": 126.62
|
| 1070 |
},
|
| 1071 |
{
|
| 1072 |
"epoch": 1.5735294117647058,
|
| 1073 |
-
"grad_norm":
|
| 1074 |
"learning_rate": 2.3774509803921568e-05,
|
| 1075 |
-
"loss":
|
| 1076 |
-
"max_memory_allocated (GB)": 2.
|
| 1077 |
-
"memory_allocated (GB)": 1.
|
| 1078 |
"step": 53500,
|
| 1079 |
"total_memory_available (GB)": 126.62
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 1.5882352941176472,
|
| 1083 |
-
"grad_norm":
|
| 1084 |
"learning_rate": 2.3529411764705884e-05,
|
| 1085 |
-
"loss":
|
| 1086 |
-
"max_memory_allocated (GB)": 2.
|
| 1087 |
-
"memory_allocated (GB)": 1.
|
| 1088 |
"step": 54000,
|
| 1089 |
"total_memory_available (GB)": 126.62
|
| 1090 |
},
|
| 1091 |
{
|
| 1092 |
"epoch": 1.6029411764705883,
|
| 1093 |
-
"grad_norm":
|
| 1094 |
"learning_rate": 2.3284313725490196e-05,
|
| 1095 |
-
"loss":
|
| 1096 |
-
"max_memory_allocated (GB)": 2.
|
| 1097 |
-
"memory_allocated (GB)": 1.
|
| 1098 |
"step": 54500,
|
| 1099 |
"total_memory_available (GB)": 126.62
|
| 1100 |
},
|
| 1101 |
{
|
| 1102 |
"epoch": 1.6176470588235294,
|
| 1103 |
-
"grad_norm":
|
| 1104 |
"learning_rate": 2.303921568627451e-05,
|
| 1105 |
-
"loss":
|
| 1106 |
-
"max_memory_allocated (GB)": 2.
|
| 1107 |
-
"memory_allocated (GB)": 1.
|
| 1108 |
"step": 55000,
|
| 1109 |
"total_memory_available (GB)": 126.62
|
| 1110 |
},
|
| 1111 |
{
|
| 1112 |
"epoch": 1.6323529411764706,
|
| 1113 |
-
"grad_norm":
|
| 1114 |
"learning_rate": 2.2794117647058825e-05,
|
| 1115 |
-
"loss":
|
| 1116 |
-
"max_memory_allocated (GB)": 2.
|
| 1117 |
-
"memory_allocated (GB)": 1.
|
| 1118 |
"step": 55500,
|
| 1119 |
"total_memory_available (GB)": 126.62
|
| 1120 |
},
|
| 1121 |
{
|
| 1122 |
"epoch": 1.6470588235294117,
|
| 1123 |
-
"grad_norm":
|
| 1124 |
"learning_rate": 2.2549019607843138e-05,
|
| 1125 |
-
"loss":
|
| 1126 |
-
"max_memory_allocated (GB)": 2.
|
| 1127 |
-
"memory_allocated (GB)": 1.
|
| 1128 |
"step": 56000,
|
| 1129 |
"total_memory_available (GB)": 126.62
|
| 1130 |
},
|
| 1131 |
{
|
| 1132 |
"epoch": 1.6617647058823528,
|
| 1133 |
-
"grad_norm":
|
| 1134 |
"learning_rate": 2.2303921568627454e-05,
|
| 1135 |
-
"loss":
|
| 1136 |
-
"max_memory_allocated (GB)": 2.
|
| 1137 |
-
"memory_allocated (GB)": 1.
|
| 1138 |
"step": 56500,
|
| 1139 |
"total_memory_available (GB)": 126.62
|
| 1140 |
},
|
| 1141 |
{
|
| 1142 |
"epoch": 1.6764705882352942,
|
| 1143 |
-
"grad_norm":
|
| 1144 |
"learning_rate": 2.2058823529411766e-05,
|
| 1145 |
-
"loss":
|
| 1146 |
-
"max_memory_allocated (GB)": 2.
|
| 1147 |
-
"memory_allocated (GB)": 1.
|
| 1148 |
"step": 57000,
|
| 1149 |
"total_memory_available (GB)": 126.62
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 1.6911764705882353,
|
| 1153 |
-
"grad_norm":
|
| 1154 |
"learning_rate": 2.181372549019608e-05,
|
| 1155 |
-
"loss":
|
| 1156 |
-
"max_memory_allocated (GB)": 2.
|
| 1157 |
-
"memory_allocated (GB)": 1.
|
| 1158 |
"step": 57500,
|
| 1159 |
"total_memory_available (GB)": 126.62
|
| 1160 |
},
|
| 1161 |
{
|
| 1162 |
"epoch": 1.7058823529411766,
|
| 1163 |
-
"grad_norm":
|
| 1164 |
"learning_rate": 2.1568627450980395e-05,
|
| 1165 |
-
"loss":
|
| 1166 |
-
"max_memory_allocated (GB)": 2.
|
| 1167 |
-
"memory_allocated (GB)": 1.
|
| 1168 |
"step": 58000,
|
| 1169 |
"total_memory_available (GB)": 126.62
|
| 1170 |
},
|
| 1171 |
{
|
| 1172 |
"epoch": 1.7205882352941178,
|
| 1173 |
-
"grad_norm":
|
| 1174 |
"learning_rate": 2.1323529411764707e-05,
|
| 1175 |
-
"loss":
|
| 1176 |
-
"max_memory_allocated (GB)": 2.
|
| 1177 |
-
"memory_allocated (GB)": 1.
|
| 1178 |
"step": 58500,
|
| 1179 |
"total_memory_available (GB)": 126.62
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 1.7352941176470589,
|
| 1183 |
-
"grad_norm":
|
| 1184 |
"learning_rate": 2.107843137254902e-05,
|
| 1185 |
-
"loss":
|
| 1186 |
-
"max_memory_allocated (GB)": 2.
|
| 1187 |
-
"memory_allocated (GB)": 1.
|
| 1188 |
"step": 59000,
|
| 1189 |
"total_memory_available (GB)": 126.62
|
| 1190 |
},
|
| 1191 |
{
|
| 1192 |
"epoch": 1.75,
|
| 1193 |
-
"grad_norm":
|
| 1194 |
"learning_rate": 2.0833333333333336e-05,
|
| 1195 |
-
"loss":
|
| 1196 |
-
"max_memory_allocated (GB)": 2.
|
| 1197 |
-
"memory_allocated (GB)": 1.
|
| 1198 |
"step": 59500,
|
| 1199 |
"total_memory_available (GB)": 126.62
|
| 1200 |
},
|
| 1201 |
{
|
| 1202 |
"epoch": 1.7647058823529411,
|
| 1203 |
-
"grad_norm":
|
| 1204 |
"learning_rate": 2.058823529411765e-05,
|
| 1205 |
-
"loss":
|
| 1206 |
-
"max_memory_allocated (GB)": 2.
|
| 1207 |
-
"memory_allocated (GB)": 1.
|
| 1208 |
"step": 60000,
|
| 1209 |
"total_memory_available (GB)": 126.62
|
| 1210 |
},
|
| 1211 |
{
|
| 1212 |
"epoch": 1.7794117647058822,
|
| 1213 |
-
"grad_norm":
|
| 1214 |
"learning_rate": 2.034313725490196e-05,
|
| 1215 |
-
"loss":
|
| 1216 |
-
"max_memory_allocated (GB)": 2.
|
| 1217 |
-
"memory_allocated (GB)": 1.
|
| 1218 |
"step": 60500,
|
| 1219 |
"total_memory_available (GB)": 126.62
|
| 1220 |
},
|
| 1221 |
{
|
| 1222 |
"epoch": 1.7941176470588234,
|
| 1223 |
-
"grad_norm":
|
| 1224 |
"learning_rate": 2.0098039215686277e-05,
|
| 1225 |
-
"loss":
|
| 1226 |
-
"max_memory_allocated (GB)": 2.
|
| 1227 |
-
"memory_allocated (GB)": 1.
|
| 1228 |
"step": 61000,
|
| 1229 |
"total_memory_available (GB)": 126.62
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
"epoch": 1.8088235294117647,
|
| 1233 |
-
"grad_norm":
|
| 1234 |
"learning_rate": 1.9852941176470586e-05,
|
| 1235 |
-
"loss":
|
| 1236 |
-
"max_memory_allocated (GB)": 2.
|
| 1237 |
-
"memory_allocated (GB)": 1.
|
| 1238 |
"step": 61500,
|
| 1239 |
"total_memory_available (GB)": 126.62
|
| 1240 |
},
|
| 1241 |
{
|
| 1242 |
"epoch": 1.8235294117647058,
|
| 1243 |
-
"grad_norm":
|
| 1244 |
"learning_rate": 1.9607843137254903e-05,
|
| 1245 |
-
"loss":
|
| 1246 |
-
"max_memory_allocated (GB)": 2.
|
| 1247 |
-
"memory_allocated (GB)": 1.
|
| 1248 |
"step": 62000,
|
| 1249 |
"total_memory_available (GB)": 126.62
|
| 1250 |
},
|
| 1251 |
{
|
| 1252 |
"epoch": 1.8382352941176472,
|
| 1253 |
-
"grad_norm":
|
| 1254 |
"learning_rate": 1.936274509803922e-05,
|
| 1255 |
-
"loss":
|
| 1256 |
-
"max_memory_allocated (GB)": 2.
|
| 1257 |
-
"memory_allocated (GB)": 1.
|
| 1258 |
"step": 62500,
|
| 1259 |
"total_memory_available (GB)": 126.62
|
| 1260 |
},
|
| 1261 |
{
|
| 1262 |
"epoch": 1.8529411764705883,
|
| 1263 |
-
"grad_norm":
|
| 1264 |
"learning_rate": 1.9117647058823528e-05,
|
| 1265 |
-
"loss":
|
| 1266 |
-
"max_memory_allocated (GB)": 2.
|
| 1267 |
-
"memory_allocated (GB)": 1.
|
| 1268 |
"step": 63000,
|
| 1269 |
"total_memory_available (GB)": 126.62
|
| 1270 |
},
|
| 1271 |
{
|
| 1272 |
"epoch": 1.8676470588235294,
|
| 1273 |
-
"grad_norm":
|
| 1274 |
"learning_rate": 1.8872549019607844e-05,
|
| 1275 |
-
"loss":
|
| 1276 |
-
"max_memory_allocated (GB)": 2.
|
| 1277 |
-
"memory_allocated (GB)": 1.
|
| 1278 |
"step": 63500,
|
| 1279 |
"total_memory_available (GB)": 126.62
|
| 1280 |
},
|
| 1281 |
{
|
| 1282 |
"epoch": 1.8823529411764706,
|
| 1283 |
-
"grad_norm":
|
| 1284 |
"learning_rate": 1.862745098039216e-05,
|
| 1285 |
-
"loss":
|
| 1286 |
-
"max_memory_allocated (GB)": 2.
|
| 1287 |
-
"memory_allocated (GB)": 1.
|
| 1288 |
"step": 64000,
|
| 1289 |
"total_memory_available (GB)": 126.62
|
| 1290 |
},
|
| 1291 |
{
|
| 1292 |
"epoch": 1.8970588235294117,
|
| 1293 |
-
"grad_norm":
|
| 1294 |
"learning_rate": 1.8382352941176472e-05,
|
| 1295 |
-
"loss":
|
| 1296 |
-
"max_memory_allocated (GB)": 2.
|
| 1297 |
-
"memory_allocated (GB)": 1.
|
| 1298 |
"step": 64500,
|
| 1299 |
"total_memory_available (GB)": 126.62
|
| 1300 |
},
|
| 1301 |
{
|
| 1302 |
"epoch": 1.9117647058823528,
|
| 1303 |
-
"grad_norm":
|
| 1304 |
"learning_rate": 1.8137254901960785e-05,
|
| 1305 |
-
"loss":
|
| 1306 |
-
"max_memory_allocated (GB)": 2.
|
| 1307 |
-
"memory_allocated (GB)": 1.
|
| 1308 |
"step": 65000,
|
| 1309 |
"total_memory_available (GB)": 126.62
|
| 1310 |
},
|
| 1311 |
{
|
| 1312 |
"epoch": 1.9264705882352942,
|
| 1313 |
-
"grad_norm":
|
| 1314 |
"learning_rate": 1.7892156862745098e-05,
|
| 1315 |
-
"loss":
|
| 1316 |
-
"max_memory_allocated (GB)": 2.
|
| 1317 |
-
"memory_allocated (GB)": 1.
|
| 1318 |
"step": 65500,
|
| 1319 |
"total_memory_available (GB)": 126.62
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 1.9411764705882353,
|
| 1323 |
-
"grad_norm":
|
| 1324 |
"learning_rate": 1.7647058823529414e-05,
|
| 1325 |
-
"loss":
|
| 1326 |
-
"max_memory_allocated (GB)": 2.
|
| 1327 |
-
"memory_allocated (GB)": 1.
|
| 1328 |
"step": 66000,
|
| 1329 |
"total_memory_available (GB)": 126.62
|
| 1330 |
},
|
| 1331 |
{
|
| 1332 |
"epoch": 1.9558823529411766,
|
| 1333 |
-
"grad_norm":
|
| 1334 |
"learning_rate": 1.7401960784313726e-05,
|
| 1335 |
-
"loss":
|
| 1336 |
-
"max_memory_allocated (GB)": 2.
|
| 1337 |
-
"memory_allocated (GB)": 1.
|
| 1338 |
"step": 66500,
|
| 1339 |
"total_memory_available (GB)": 126.62
|
| 1340 |
},
|
| 1341 |
{
|
| 1342 |
"epoch": 1.9705882352941178,
|
| 1343 |
-
"grad_norm":
|
| 1344 |
"learning_rate": 1.715686274509804e-05,
|
| 1345 |
-
"loss":
|
| 1346 |
-
"max_memory_allocated (GB)": 2.
|
| 1347 |
-
"memory_allocated (GB)": 1.
|
| 1348 |
"step": 67000,
|
| 1349 |
"total_memory_available (GB)": 126.62
|
| 1350 |
},
|
| 1351 |
{
|
| 1352 |
"epoch": 1.9852941176470589,
|
| 1353 |
-
"grad_norm":
|
| 1354 |
"learning_rate": 1.6911764705882355e-05,
|
| 1355 |
-
"loss":
|
| 1356 |
-
"max_memory_allocated (GB)": 2.
|
| 1357 |
-
"memory_allocated (GB)": 1.
|
| 1358 |
"step": 67500,
|
| 1359 |
"total_memory_available (GB)": 126.62
|
| 1360 |
},
|
| 1361 |
{
|
| 1362 |
"epoch": 2.0,
|
| 1363 |
-
"grad_norm":
|
| 1364 |
"learning_rate": 1.6666666666666667e-05,
|
| 1365 |
-
"loss":
|
| 1366 |
-
"max_memory_allocated (GB)": 2.
|
| 1367 |
-
"memory_allocated (GB)": 1.
|
| 1368 |
"step": 68000,
|
| 1369 |
"total_memory_available (GB)": 126.62
|
| 1370 |
},
|
| 1371 |
{
|
| 1372 |
"epoch": 2.014705882352941,
|
| 1373 |
-
"grad_norm":
|
| 1374 |
"learning_rate": 1.642156862745098e-05,
|
| 1375 |
-
"loss":
|
| 1376 |
-
"max_memory_allocated (GB)": 2.
|
| 1377 |
-
"memory_allocated (GB)": 1.
|
| 1378 |
"step": 68500,
|
| 1379 |
"total_memory_available (GB)": 126.62
|
| 1380 |
},
|
| 1381 |
{
|
| 1382 |
"epoch": 2.0294117647058822,
|
| 1383 |
-
"grad_norm":
|
| 1384 |
"learning_rate": 1.6176470588235296e-05,
|
| 1385 |
-
"loss":
|
| 1386 |
-
"max_memory_allocated (GB)": 2.
|
| 1387 |
-
"memory_allocated (GB)": 1.
|
| 1388 |
"step": 69000,
|
| 1389 |
"total_memory_available (GB)": 126.62
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 2.0441176470588234,
|
| 1393 |
-
"grad_norm":
|
| 1394 |
"learning_rate": 1.593137254901961e-05,
|
| 1395 |
-
"loss":
|
| 1396 |
-
"max_memory_allocated (GB)": 2.
|
| 1397 |
-
"memory_allocated (GB)": 1.
|
| 1398 |
"step": 69500,
|
| 1399 |
"total_memory_available (GB)": 126.62
|
| 1400 |
},
|
| 1401 |
{
|
| 1402 |
"epoch": 2.0588235294117645,
|
| 1403 |
-
"grad_norm":
|
| 1404 |
"learning_rate": 1.568627450980392e-05,
|
| 1405 |
-
"loss":
|
| 1406 |
-
"max_memory_allocated (GB)": 2.
|
| 1407 |
-
"memory_allocated (GB)": 1.
|
| 1408 |
"step": 70000,
|
| 1409 |
"total_memory_available (GB)": 126.62
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
"epoch": 2.073529411764706,
|
| 1413 |
-
"grad_norm":
|
| 1414 |
"learning_rate": 1.5441176470588237e-05,
|
| 1415 |
-
"loss":
|
| 1416 |
-
"max_memory_allocated (GB)": 2.
|
| 1417 |
-
"memory_allocated (GB)": 1.
|
| 1418 |
"step": 70500,
|
| 1419 |
"total_memory_available (GB)": 126.62
|
| 1420 |
},
|
| 1421 |
{
|
| 1422 |
"epoch": 2.088235294117647,
|
| 1423 |
-
"grad_norm":
|
| 1424 |
"learning_rate": 1.5196078431372548e-05,
|
| 1425 |
-
"loss":
|
| 1426 |
-
"max_memory_allocated (GB)": 2.
|
| 1427 |
-
"memory_allocated (GB)": 1.
|
| 1428 |
"step": 71000,
|
| 1429 |
"total_memory_available (GB)": 126.62
|
| 1430 |
},
|
| 1431 |
{
|
| 1432 |
"epoch": 2.1029411764705883,
|
| 1433 |
-
"grad_norm":
|
| 1434 |
"learning_rate": 1.4950980392156863e-05,
|
| 1435 |
-
"loss":
|
| 1436 |
-
"max_memory_allocated (GB)": 2.
|
| 1437 |
-
"memory_allocated (GB)": 1.
|
| 1438 |
"step": 71500,
|
| 1439 |
"total_memory_available (GB)": 126.62
|
| 1440 |
},
|
| 1441 |
{
|
| 1442 |
"epoch": 2.1176470588235294,
|
| 1443 |
-
"grad_norm":
|
| 1444 |
"learning_rate": 1.4705882352941177e-05,
|
| 1445 |
-
"loss":
|
| 1446 |
-
"max_memory_allocated (GB)": 2.
|
| 1447 |
-
"memory_allocated (GB)": 1.
|
| 1448 |
"step": 72000,
|
| 1449 |
"total_memory_available (GB)": 126.62
|
| 1450 |
},
|
| 1451 |
{
|
| 1452 |
"epoch": 2.1323529411764706,
|
| 1453 |
-
"grad_norm":
|
| 1454 |
"learning_rate": 1.4460784313725493e-05,
|
| 1455 |
-
"loss":
|
| 1456 |
-
"max_memory_allocated (GB)": 2.
|
| 1457 |
-
"memory_allocated (GB)": 1.
|
| 1458 |
"step": 72500,
|
| 1459 |
"total_memory_available (GB)": 126.62
|
| 1460 |
},
|
| 1461 |
{
|
| 1462 |
"epoch": 2.1470588235294117,
|
| 1463 |
-
"grad_norm":
|
| 1464 |
"learning_rate": 1.4215686274509804e-05,
|
| 1465 |
-
"loss":
|
| 1466 |
-
"max_memory_allocated (GB)": 2.
|
| 1467 |
-
"memory_allocated (GB)": 1.
|
| 1468 |
"step": 73000,
|
| 1469 |
"total_memory_available (GB)": 126.62
|
| 1470 |
},
|
| 1471 |
{
|
| 1472 |
"epoch": 2.161764705882353,
|
| 1473 |
-
"grad_norm":
|
| 1474 |
"learning_rate": 1.3970588235294118e-05,
|
| 1475 |
-
"loss":
|
| 1476 |
-
"max_memory_allocated (GB)": 2.
|
| 1477 |
-
"memory_allocated (GB)": 1.
|
| 1478 |
"step": 73500,
|
| 1479 |
"total_memory_available (GB)": 126.62
|
| 1480 |
},
|
| 1481 |
{
|
| 1482 |
"epoch": 2.176470588235294,
|
| 1483 |
-
"grad_norm":
|
| 1484 |
"learning_rate": 1.3725490196078432e-05,
|
| 1485 |
-
"loss":
|
| 1486 |
-
"max_memory_allocated (GB)": 2.
|
| 1487 |
-
"memory_allocated (GB)": 1.
|
| 1488 |
"step": 74000,
|
| 1489 |
"total_memory_available (GB)": 126.62
|
| 1490 |
},
|
| 1491 |
{
|
| 1492 |
"epoch": 2.1911764705882355,
|
| 1493 |
-
"grad_norm":
|
| 1494 |
"learning_rate": 1.3480392156862745e-05,
|
| 1495 |
-
"loss":
|
| 1496 |
-
"max_memory_allocated (GB)": 2.
|
| 1497 |
-
"memory_allocated (GB)": 1.
|
| 1498 |
"step": 74500,
|
| 1499 |
"total_memory_available (GB)": 126.62
|
| 1500 |
},
|
| 1501 |
{
|
| 1502 |
"epoch": 2.2058823529411766,
|
| 1503 |
-
"grad_norm":
|
| 1504 |
"learning_rate": 1.323529411764706e-05,
|
| 1505 |
-
"loss":
|
| 1506 |
-
"max_memory_allocated (GB)": 2.
|
| 1507 |
-
"memory_allocated (GB)": 1.
|
| 1508 |
"step": 75000,
|
| 1509 |
"total_memory_available (GB)": 126.62
|
| 1510 |
},
|
| 1511 |
{
|
| 1512 |
"epoch": 2.2205882352941178,
|
| 1513 |
-
"grad_norm":
|
| 1514 |
"learning_rate": 1.2990196078431374e-05,
|
| 1515 |
-
"loss":
|
| 1516 |
-
"max_memory_allocated (GB)": 2.
|
| 1517 |
-
"memory_allocated (GB)": 1.
|
| 1518 |
"step": 75500,
|
| 1519 |
"total_memory_available (GB)": 126.62
|
| 1520 |
},
|
| 1521 |
{
|
| 1522 |
"epoch": 2.235294117647059,
|
| 1523 |
-
"grad_norm":
|
| 1524 |
"learning_rate": 1.2745098039215686e-05,
|
| 1525 |
-
"loss":
|
| 1526 |
-
"max_memory_allocated (GB)": 2.
|
| 1527 |
-
"memory_allocated (GB)": 1.
|
| 1528 |
"step": 76000,
|
| 1529 |
"total_memory_available (GB)": 126.62
|
| 1530 |
},
|
| 1531 |
{
|
| 1532 |
"epoch": 2.25,
|
| 1533 |
-
"grad_norm":
|
| 1534 |
"learning_rate": 1.25e-05,
|
| 1535 |
-
"loss":
|
| 1536 |
-
"max_memory_allocated (GB)": 2.
|
| 1537 |
-
"memory_allocated (GB)": 1.
|
| 1538 |
"step": 76500,
|
| 1539 |
"total_memory_available (GB)": 126.62
|
| 1540 |
},
|
| 1541 |
{
|
| 1542 |
"epoch": 2.264705882352941,
|
| 1543 |
-
"grad_norm":
|
| 1544 |
"learning_rate": 1.2254901960784313e-05,
|
| 1545 |
-
"loss":
|
| 1546 |
-
"max_memory_allocated (GB)": 2.
|
| 1547 |
-
"memory_allocated (GB)": 1.
|
| 1548 |
"step": 77000,
|
| 1549 |
"total_memory_available (GB)": 126.62
|
| 1550 |
},
|
| 1551 |
{
|
| 1552 |
"epoch": 2.2794117647058822,
|
| 1553 |
-
"grad_norm":
|
| 1554 |
"learning_rate": 1.200980392156863e-05,
|
| 1555 |
-
"loss":
|
| 1556 |
-
"max_memory_allocated (GB)": 2.
|
| 1557 |
-
"memory_allocated (GB)": 1.
|
| 1558 |
"step": 77500,
|
| 1559 |
"total_memory_available (GB)": 126.62
|
| 1560 |
},
|
| 1561 |
{
|
| 1562 |
"epoch": 2.2941176470588234,
|
| 1563 |
-
"grad_norm":
|
| 1564 |
"learning_rate": 1.1764705882352942e-05,
|
| 1565 |
-
"loss":
|
| 1566 |
-
"max_memory_allocated (GB)": 2.
|
| 1567 |
-
"memory_allocated (GB)": 1.
|
| 1568 |
"step": 78000,
|
| 1569 |
"total_memory_available (GB)": 126.62
|
| 1570 |
},
|
| 1571 |
{
|
| 1572 |
"epoch": 2.3088235294117645,
|
| 1573 |
-
"grad_norm":
|
| 1574 |
"learning_rate": 1.1519607843137254e-05,
|
| 1575 |
-
"loss":
|
| 1576 |
-
"max_memory_allocated (GB)": 2.
|
| 1577 |
-
"memory_allocated (GB)": 1.
|
| 1578 |
"step": 78500,
|
| 1579 |
"total_memory_available (GB)": 126.62
|
| 1580 |
},
|
| 1581 |
{
|
| 1582 |
"epoch": 2.323529411764706,
|
| 1583 |
-
"grad_norm":
|
| 1584 |
"learning_rate": 1.1274509803921569e-05,
|
| 1585 |
-
"loss":
|
| 1586 |
-
"max_memory_allocated (GB)": 2.
|
| 1587 |
-
"memory_allocated (GB)": 1.
|
| 1588 |
"step": 79000,
|
| 1589 |
"total_memory_available (GB)": 126.62
|
| 1590 |
},
|
| 1591 |
{
|
| 1592 |
"epoch": 2.338235294117647,
|
| 1593 |
-
"grad_norm":
|
| 1594 |
"learning_rate": 1.1029411764705883e-05,
|
| 1595 |
-
"loss":
|
| 1596 |
-
"max_memory_allocated (GB)": 2.
|
| 1597 |
-
"memory_allocated (GB)": 1.
|
| 1598 |
"step": 79500,
|
| 1599 |
"total_memory_available (GB)": 126.62
|
| 1600 |
},
|
| 1601 |
{
|
| 1602 |
"epoch": 2.3529411764705883,
|
| 1603 |
-
"grad_norm":
|
| 1604 |
"learning_rate": 1.0784313725490197e-05,
|
| 1605 |
-
"loss":
|
| 1606 |
-
"max_memory_allocated (GB)": 2.
|
| 1607 |
-
"memory_allocated (GB)": 1.
|
| 1608 |
"step": 80000,
|
| 1609 |
"total_memory_available (GB)": 126.62
|
| 1610 |
},
|
| 1611 |
{
|
| 1612 |
"epoch": 2.3676470588235294,
|
| 1613 |
-
"grad_norm":
|
| 1614 |
"learning_rate": 1.053921568627451e-05,
|
| 1615 |
-
"loss":
|
| 1616 |
-
"max_memory_allocated (GB)": 2.
|
| 1617 |
-
"memory_allocated (GB)": 1.
|
| 1618 |
"step": 80500,
|
| 1619 |
"total_memory_available (GB)": 126.62
|
| 1620 |
},
|
| 1621 |
{
|
| 1622 |
"epoch": 2.3823529411764706,
|
| 1623 |
-
"grad_norm":
|
| 1624 |
"learning_rate": 1.0294117647058824e-05,
|
| 1625 |
-
"loss":
|
| 1626 |
-
"max_memory_allocated (GB)": 2.
|
| 1627 |
-
"memory_allocated (GB)": 1.
|
| 1628 |
"step": 81000,
|
| 1629 |
"total_memory_available (GB)": 126.62
|
| 1630 |
},
|
| 1631 |
{
|
| 1632 |
"epoch": 2.3970588235294117,
|
| 1633 |
-
"grad_norm":
|
| 1634 |
"learning_rate": 1.0049019607843139e-05,
|
| 1635 |
-
"loss":
|
| 1636 |
-
"max_memory_allocated (GB)": 2.
|
| 1637 |
-
"memory_allocated (GB)": 1.
|
| 1638 |
"step": 81500,
|
| 1639 |
"total_memory_available (GB)": 126.62
|
| 1640 |
},
|
| 1641 |
{
|
| 1642 |
"epoch": 2.411764705882353,
|
| 1643 |
-
"grad_norm":
|
| 1644 |
"learning_rate": 9.803921568627451e-06,
|
| 1645 |
-
"loss":
|
| 1646 |
-
"max_memory_allocated (GB)": 2.
|
| 1647 |
-
"memory_allocated (GB)": 1.
|
| 1648 |
"step": 82000,
|
| 1649 |
"total_memory_available (GB)": 126.62
|
| 1650 |
},
|
| 1651 |
{
|
| 1652 |
"epoch": 2.426470588235294,
|
| 1653 |
-
"grad_norm":
|
| 1654 |
"learning_rate": 9.558823529411764e-06,
|
| 1655 |
-
"loss":
|
| 1656 |
-
"max_memory_allocated (GB)": 2.
|
| 1657 |
-
"memory_allocated (GB)": 1.
|
| 1658 |
"step": 82500,
|
| 1659 |
"total_memory_available (GB)": 126.62
|
| 1660 |
},
|
| 1661 |
{
|
| 1662 |
"epoch": 2.4411764705882355,
|
| 1663 |
-
"grad_norm":
|
| 1664 |
"learning_rate": 9.31372549019608e-06,
|
| 1665 |
-
"loss":
|
| 1666 |
-
"max_memory_allocated (GB)": 2.
|
| 1667 |
-
"memory_allocated (GB)": 1.
|
| 1668 |
"step": 83000,
|
| 1669 |
"total_memory_available (GB)": 126.62
|
| 1670 |
},
|
| 1671 |
{
|
| 1672 |
"epoch": 2.4558823529411766,
|
| 1673 |
-
"grad_norm":
|
| 1674 |
"learning_rate": 9.068627450980392e-06,
|
| 1675 |
-
"loss":
|
| 1676 |
-
"max_memory_allocated (GB)": 2.
|
| 1677 |
-
"memory_allocated (GB)": 1.
|
| 1678 |
"step": 83500,
|
| 1679 |
"total_memory_available (GB)": 126.62
|
| 1680 |
},
|
| 1681 |
{
|
| 1682 |
"epoch": 2.4705882352941178,
|
| 1683 |
-
"grad_norm":
|
| 1684 |
"learning_rate": 8.823529411764707e-06,
|
| 1685 |
-
"loss":
|
| 1686 |
-
"max_memory_allocated (GB)": 2.
|
| 1687 |
-
"memory_allocated (GB)": 1.
|
| 1688 |
"step": 84000,
|
| 1689 |
"total_memory_available (GB)": 126.62
|
| 1690 |
},
|
| 1691 |
{
|
| 1692 |
"epoch": 2.485294117647059,
|
| 1693 |
-
"grad_norm":
|
| 1694 |
"learning_rate": 8.57843137254902e-06,
|
| 1695 |
-
"loss":
|
| 1696 |
-
"max_memory_allocated (GB)": 2.
|
| 1697 |
-
"memory_allocated (GB)": 1.
|
| 1698 |
"step": 84500,
|
| 1699 |
"total_memory_available (GB)": 126.62
|
| 1700 |
},
|
| 1701 |
{
|
| 1702 |
"epoch": 2.5,
|
| 1703 |
-
"grad_norm":
|
| 1704 |
"learning_rate": 8.333333333333334e-06,
|
| 1705 |
-
"loss":
|
| 1706 |
-
"max_memory_allocated (GB)": 2.
|
| 1707 |
-
"memory_allocated (GB)": 1.
|
| 1708 |
"step": 85000,
|
| 1709 |
"total_memory_available (GB)": 126.62
|
| 1710 |
},
|
| 1711 |
{
|
| 1712 |
"epoch": 2.514705882352941,
|
| 1713 |
-
"grad_norm":
|
| 1714 |
"learning_rate": 8.088235294117648e-06,
|
| 1715 |
-
"loss":
|
| 1716 |
-
"max_memory_allocated (GB)": 2.
|
| 1717 |
-
"memory_allocated (GB)": 1.
|
| 1718 |
"step": 85500,
|
| 1719 |
"total_memory_available (GB)": 126.62
|
| 1720 |
},
|
| 1721 |
{
|
| 1722 |
"epoch": 2.5294117647058822,
|
| 1723 |
-
"grad_norm":
|
| 1724 |
"learning_rate": 7.84313725490196e-06,
|
| 1725 |
-
"loss":
|
| 1726 |
-
"max_memory_allocated (GB)": 2.
|
| 1727 |
-
"memory_allocated (GB)": 1.
|
| 1728 |
"step": 86000,
|
| 1729 |
"total_memory_available (GB)": 126.62
|
| 1730 |
},
|
| 1731 |
{
|
| 1732 |
"epoch": 2.5441176470588234,
|
| 1733 |
-
"grad_norm":
|
| 1734 |
"learning_rate": 7.598039215686274e-06,
|
| 1735 |
-
"loss":
|
| 1736 |
-
"max_memory_allocated (GB)": 2.
|
| 1737 |
-
"memory_allocated (GB)": 1.
|
| 1738 |
"step": 86500,
|
| 1739 |
"total_memory_available (GB)": 126.62
|
| 1740 |
},
|
| 1741 |
{
|
| 1742 |
"epoch": 2.5588235294117645,
|
| 1743 |
-
"grad_norm":
|
| 1744 |
"learning_rate": 7.3529411764705884e-06,
|
| 1745 |
-
"loss":
|
| 1746 |
-
"max_memory_allocated (GB)": 2.
|
| 1747 |
-
"memory_allocated (GB)": 1.
|
| 1748 |
"step": 87000,
|
| 1749 |
"total_memory_available (GB)": 126.62
|
| 1750 |
},
|
| 1751 |
{
|
| 1752 |
"epoch": 2.5735294117647056,
|
| 1753 |
-
"grad_norm":
|
| 1754 |
"learning_rate": 7.107843137254902e-06,
|
| 1755 |
-
"loss":
|
| 1756 |
-
"max_memory_allocated (GB)": 2.
|
| 1757 |
-
"memory_allocated (GB)": 1.
|
| 1758 |
"step": 87500,
|
| 1759 |
"total_memory_available (GB)": 126.62
|
| 1760 |
},
|
| 1761 |
{
|
| 1762 |
"epoch": 2.588235294117647,
|
| 1763 |
-
"grad_norm":
|
| 1764 |
"learning_rate": 6.862745098039216e-06,
|
| 1765 |
-
"loss":
|
| 1766 |
-
"max_memory_allocated (GB)": 2.
|
| 1767 |
-
"memory_allocated (GB)": 1.
|
| 1768 |
"step": 88000,
|
| 1769 |
"total_memory_available (GB)": 126.62
|
| 1770 |
},
|
| 1771 |
{
|
| 1772 |
"epoch": 2.6029411764705883,
|
| 1773 |
-
"grad_norm":
|
| 1774 |
"learning_rate": 6.61764705882353e-06,
|
| 1775 |
-
"loss":
|
| 1776 |
-
"max_memory_allocated (GB)": 2.
|
| 1777 |
-
"memory_allocated (GB)": 1.
|
| 1778 |
"step": 88500,
|
| 1779 |
"total_memory_available (GB)": 126.62
|
| 1780 |
},
|
| 1781 |
{
|
| 1782 |
"epoch": 2.6176470588235294,
|
| 1783 |
-
"grad_norm":
|
| 1784 |
"learning_rate": 6.372549019607843e-06,
|
| 1785 |
-
"loss":
|
| 1786 |
-
"max_memory_allocated (GB)": 2.
|
| 1787 |
-
"memory_allocated (GB)": 1.
|
| 1788 |
"step": 89000,
|
| 1789 |
"total_memory_available (GB)": 126.62
|
| 1790 |
},
|
| 1791 |
{
|
| 1792 |
"epoch": 2.6323529411764706,
|
| 1793 |
-
"grad_norm":
|
| 1794 |
"learning_rate": 6.127450980392157e-06,
|
| 1795 |
-
"loss":
|
| 1796 |
-
"max_memory_allocated (GB)": 2.
|
| 1797 |
-
"memory_allocated (GB)": 1.
|
| 1798 |
"step": 89500,
|
| 1799 |
"total_memory_available (GB)": 126.62
|
| 1800 |
},
|
| 1801 |
{
|
| 1802 |
"epoch": 2.6470588235294117,
|
| 1803 |
-
"grad_norm":
|
| 1804 |
"learning_rate": 5.882352941176471e-06,
|
| 1805 |
-
"loss":
|
| 1806 |
-
"max_memory_allocated (GB)": 2.
|
| 1807 |
-
"memory_allocated (GB)": 1.
|
| 1808 |
"step": 90000,
|
| 1809 |
"total_memory_available (GB)": 126.62
|
| 1810 |
},
|
| 1811 |
{
|
| 1812 |
"epoch": 2.661764705882353,
|
| 1813 |
-
"grad_norm":
|
| 1814 |
"learning_rate": 5.637254901960784e-06,
|
| 1815 |
-
"loss":
|
| 1816 |
-
"max_memory_allocated (GB)": 2.
|
| 1817 |
-
"memory_allocated (GB)": 1.
|
| 1818 |
"step": 90500,
|
| 1819 |
"total_memory_available (GB)": 126.62
|
| 1820 |
},
|
| 1821 |
{
|
| 1822 |
"epoch": 2.6764705882352944,
|
| 1823 |
-
"grad_norm":
|
| 1824 |
"learning_rate": 5.392156862745099e-06,
|
| 1825 |
-
"loss":
|
| 1826 |
-
"max_memory_allocated (GB)": 2.
|
| 1827 |
-
"memory_allocated (GB)": 1.
|
| 1828 |
"step": 91000,
|
| 1829 |
"total_memory_available (GB)": 126.62
|
| 1830 |
},
|
| 1831 |
{
|
| 1832 |
"epoch": 2.6911764705882355,
|
| 1833 |
-
"grad_norm":
|
| 1834 |
"learning_rate": 5.147058823529412e-06,
|
| 1835 |
-
"loss":
|
| 1836 |
-
"max_memory_allocated (GB)": 2.
|
| 1837 |
-
"memory_allocated (GB)": 1.
|
| 1838 |
"step": 91500,
|
| 1839 |
"total_memory_available (GB)": 126.62
|
| 1840 |
},
|
| 1841 |
{
|
| 1842 |
"epoch": 2.7058823529411766,
|
| 1843 |
-
"grad_norm":
|
| 1844 |
"learning_rate": 4.901960784313726e-06,
|
| 1845 |
-
"loss":
|
| 1846 |
-
"max_memory_allocated (GB)": 2.
|
| 1847 |
-
"memory_allocated (GB)": 1.
|
| 1848 |
"step": 92000,
|
| 1849 |
"total_memory_available (GB)": 126.62
|
| 1850 |
},
|
| 1851 |
{
|
| 1852 |
"epoch": 2.7205882352941178,
|
| 1853 |
-
"grad_norm":
|
| 1854 |
"learning_rate": 4.65686274509804e-06,
|
| 1855 |
-
"loss":
|
| 1856 |
-
"max_memory_allocated (GB)": 2.
|
| 1857 |
-
"memory_allocated (GB)": 1.
|
| 1858 |
"step": 92500,
|
| 1859 |
"total_memory_available (GB)": 126.62
|
| 1860 |
},
|
| 1861 |
{
|
| 1862 |
"epoch": 2.735294117647059,
|
| 1863 |
-
"grad_norm":
|
| 1864 |
"learning_rate": 4.411764705882353e-06,
|
| 1865 |
-
"loss":
|
| 1866 |
-
"max_memory_allocated (GB)": 2.
|
| 1867 |
-
"memory_allocated (GB)": 1.
|
| 1868 |
"step": 93000,
|
| 1869 |
"total_memory_available (GB)": 126.62
|
| 1870 |
},
|
| 1871 |
{
|
| 1872 |
"epoch": 2.75,
|
| 1873 |
-
"grad_norm":
|
| 1874 |
"learning_rate": 4.166666666666667e-06,
|
| 1875 |
-
"loss":
|
| 1876 |
-
"max_memory_allocated (GB)": 2.
|
| 1877 |
-
"memory_allocated (GB)": 1.
|
| 1878 |
"step": 93500,
|
| 1879 |
"total_memory_available (GB)": 126.62
|
| 1880 |
},
|
| 1881 |
{
|
| 1882 |
"epoch": 2.764705882352941,
|
| 1883 |
-
"grad_norm":
|
| 1884 |
"learning_rate": 3.92156862745098e-06,
|
| 1885 |
-
"loss":
|
| 1886 |
-
"max_memory_allocated (GB)": 2.
|
| 1887 |
-
"memory_allocated (GB)": 1.
|
| 1888 |
"step": 94000,
|
| 1889 |
"total_memory_available (GB)": 126.62
|
| 1890 |
},
|
| 1891 |
{
|
| 1892 |
"epoch": 2.7794117647058822,
|
| 1893 |
-
"grad_norm":
|
| 1894 |
"learning_rate": 3.6764705882352942e-06,
|
| 1895 |
-
"loss":
|
| 1896 |
-
"max_memory_allocated (GB)": 2.
|
| 1897 |
-
"memory_allocated (GB)": 1.
|
| 1898 |
"step": 94500,
|
| 1899 |
"total_memory_available (GB)": 126.62
|
| 1900 |
},
|
| 1901 |
{
|
| 1902 |
"epoch": 2.7941176470588234,
|
| 1903 |
-
"grad_norm":
|
| 1904 |
"learning_rate": 3.431372549019608e-06,
|
| 1905 |
-
"loss":
|
| 1906 |
-
"max_memory_allocated (GB)": 2.
|
| 1907 |
-
"memory_allocated (GB)": 1.
|
| 1908 |
"step": 95000,
|
| 1909 |
"total_memory_available (GB)": 126.62
|
| 1910 |
},
|
| 1911 |
{
|
| 1912 |
"epoch": 2.8088235294117645,
|
| 1913 |
-
"grad_norm":
|
| 1914 |
"learning_rate": 3.1862745098039216e-06,
|
| 1915 |
-
"loss":
|
| 1916 |
-
"max_memory_allocated (GB)": 2.
|
| 1917 |
-
"memory_allocated (GB)": 1.
|
| 1918 |
"step": 95500,
|
| 1919 |
"total_memory_available (GB)": 126.62
|
| 1920 |
},
|
| 1921 |
{
|
| 1922 |
"epoch": 2.8235294117647056,
|
| 1923 |
-
"grad_norm":
|
| 1924 |
"learning_rate": 2.9411764705882355e-06,
|
| 1925 |
-
"loss":
|
| 1926 |
-
"max_memory_allocated (GB)": 2.
|
| 1927 |
-
"memory_allocated (GB)": 1.
|
| 1928 |
"step": 96000,
|
| 1929 |
"total_memory_available (GB)": 126.62
|
| 1930 |
},
|
| 1931 |
{
|
| 1932 |
"epoch": 2.838235294117647,
|
| 1933 |
-
"grad_norm":
|
| 1934 |
"learning_rate": 2.6960784313725493e-06,
|
| 1935 |
-
"loss":
|
| 1936 |
-
"max_memory_allocated (GB)": 2.
|
| 1937 |
-
"memory_allocated (GB)": 1.
|
| 1938 |
"step": 96500,
|
| 1939 |
"total_memory_available (GB)": 126.62
|
| 1940 |
},
|
| 1941 |
{
|
| 1942 |
"epoch": 2.8529411764705883,
|
| 1943 |
-
"grad_norm":
|
| 1944 |
"learning_rate": 2.450980392156863e-06,
|
| 1945 |
-
"loss":
|
| 1946 |
-
"max_memory_allocated (GB)": 2.
|
| 1947 |
-
"memory_allocated (GB)": 1.
|
| 1948 |
"step": 97000,
|
| 1949 |
"total_memory_available (GB)": 126.62
|
| 1950 |
},
|
| 1951 |
{
|
| 1952 |
"epoch": 2.8676470588235294,
|
| 1953 |
-
"grad_norm":
|
| 1954 |
"learning_rate": 2.2058823529411767e-06,
|
| 1955 |
-
"loss":
|
| 1956 |
-
"max_memory_allocated (GB)": 2.
|
| 1957 |
-
"memory_allocated (GB)": 1.
|
| 1958 |
"step": 97500,
|
| 1959 |
"total_memory_available (GB)": 126.62
|
| 1960 |
},
|
| 1961 |
{
|
| 1962 |
"epoch": 2.8823529411764706,
|
| 1963 |
-
"grad_norm":
|
| 1964 |
"learning_rate": 1.96078431372549e-06,
|
| 1965 |
-
"loss":
|
| 1966 |
-
"max_memory_allocated (GB)": 2.
|
| 1967 |
-
"memory_allocated (GB)": 1.
|
| 1968 |
"step": 98000,
|
| 1969 |
"total_memory_available (GB)": 126.62
|
| 1970 |
},
|
| 1971 |
{
|
| 1972 |
"epoch": 2.8970588235294117,
|
| 1973 |
-
"grad_norm":
|
| 1974 |
"learning_rate": 1.715686274509804e-06,
|
| 1975 |
-
"loss":
|
| 1976 |
-
"max_memory_allocated (GB)": 2.
|
| 1977 |
-
"memory_allocated (GB)": 1.
|
| 1978 |
"step": 98500,
|
| 1979 |
"total_memory_available (GB)": 126.62
|
| 1980 |
},
|
| 1981 |
{
|
| 1982 |
"epoch": 2.911764705882353,
|
| 1983 |
-
"grad_norm":
|
| 1984 |
"learning_rate": 1.4705882352941177e-06,
|
| 1985 |
-
"loss":
|
| 1986 |
-
"max_memory_allocated (GB)": 2.
|
| 1987 |
-
"memory_allocated (GB)": 1.
|
| 1988 |
"step": 99000,
|
| 1989 |
"total_memory_available (GB)": 126.62
|
| 1990 |
},
|
| 1991 |
{
|
| 1992 |
"epoch": 2.9264705882352944,
|
| 1993 |
-
"grad_norm":
|
| 1994 |
"learning_rate": 1.2254901960784314e-06,
|
| 1995 |
-
"loss":
|
| 1996 |
-
"max_memory_allocated (GB)": 2.
|
| 1997 |
-
"memory_allocated (GB)": 1.
|
| 1998 |
"step": 99500,
|
| 1999 |
"total_memory_available (GB)": 126.62
|
| 2000 |
},
|
| 2001 |
{
|
| 2002 |
"epoch": 2.9411764705882355,
|
| 2003 |
-
"grad_norm":
|
| 2004 |
"learning_rate": 9.80392156862745e-07,
|
| 2005 |
-
"loss":
|
| 2006 |
-
"max_memory_allocated (GB)": 2.
|
| 2007 |
-
"memory_allocated (GB)": 1.
|
| 2008 |
"step": 100000,
|
| 2009 |
"total_memory_available (GB)": 126.62
|
| 2010 |
},
|
| 2011 |
{
|
| 2012 |
"epoch": 2.9558823529411766,
|
| 2013 |
-
"grad_norm":
|
| 2014 |
"learning_rate": 7.352941176470589e-07,
|
| 2015 |
-
"loss":
|
| 2016 |
-
"max_memory_allocated (GB)": 2.
|
| 2017 |
-
"memory_allocated (GB)": 1.
|
| 2018 |
"step": 100500,
|
| 2019 |
"total_memory_available (GB)": 126.62
|
| 2020 |
},
|
| 2021 |
{
|
| 2022 |
"epoch": 2.9705882352941178,
|
| 2023 |
-
"grad_norm":
|
| 2024 |
"learning_rate": 4.901960784313725e-07,
|
| 2025 |
-
"loss":
|
| 2026 |
-
"max_memory_allocated (GB)": 2.
|
| 2027 |
-
"memory_allocated (GB)": 1.
|
| 2028 |
"step": 101000,
|
| 2029 |
"total_memory_available (GB)": 126.62
|
| 2030 |
},
|
| 2031 |
{
|
| 2032 |
"epoch": 2.985294117647059,
|
| 2033 |
-
"grad_norm":
|
| 2034 |
"learning_rate": 2.4509803921568627e-07,
|
| 2035 |
-
"loss":
|
| 2036 |
-
"max_memory_allocated (GB)": 2.
|
| 2037 |
-
"memory_allocated (GB)": 1.
|
| 2038 |
"step": 101500,
|
| 2039 |
"total_memory_available (GB)": 126.62
|
| 2040 |
},
|
| 2041 |
{
|
| 2042 |
"epoch": 3.0,
|
| 2043 |
-
"grad_norm":
|
| 2044 |
"learning_rate": 0.0,
|
| 2045 |
-
"loss":
|
| 2046 |
-
"max_memory_allocated (GB)": 2.
|
| 2047 |
-
"memory_allocated (GB)": 1.
|
| 2048 |
"step": 102000,
|
| 2049 |
"total_memory_available (GB)": 126.62
|
| 2050 |
},
|
| 2051 |
{
|
| 2052 |
"epoch": 3.0,
|
| 2053 |
-
"max_memory_allocated (GB)":
|
| 2054 |
-
"memory_allocated (GB)":
|
| 2055 |
"step": 102000,
|
| 2056 |
"total_flos": 6.324139790696448e+19,
|
| 2057 |
"total_memory_available (GB)": 126.62,
|
| 2058 |
-
"train_loss": 0.
|
| 2059 |
-
"train_runtime":
|
| 2060 |
-
"train_samples_per_second":
|
| 2061 |
-
"train_steps_per_second":
|
| 2062 |
}
|
| 2063 |
],
|
| 2064 |
"logging_steps": 500,
|
|
|
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.014705882352941176,
|
| 13 |
+
"grad_norm": 3.0940372943878174,
|
| 14 |
"learning_rate": 4.975490196078432e-05,
|
| 15 |
+
"loss": 2.042,
|
| 16 |
+
"max_memory_allocated (GB)": 2.1,
|
| 17 |
+
"memory_allocated (GB)": 1.46,
|
| 18 |
"step": 500,
|
| 19 |
"total_memory_available (GB)": 126.62
|
| 20 |
},
|
| 21 |
{
|
| 22 |
"epoch": 0.029411764705882353,
|
| 23 |
+
"grad_norm": 3.2075328826904297,
|
| 24 |
"learning_rate": 4.9509803921568634e-05,
|
| 25 |
+
"loss": 1.5982,
|
| 26 |
+
"max_memory_allocated (GB)": 2.1,
|
| 27 |
+
"memory_allocated (GB)": 1.46,
|
| 28 |
"step": 1000,
|
| 29 |
"total_memory_available (GB)": 126.62
|
| 30 |
},
|
| 31 |
{
|
| 32 |
"epoch": 0.04411764705882353,
|
| 33 |
+
"grad_norm": 4.371872425079346,
|
| 34 |
"learning_rate": 4.9264705882352944e-05,
|
| 35 |
+
"loss": 1.4943,
|
| 36 |
+
"max_memory_allocated (GB)": 2.1,
|
| 37 |
+
"memory_allocated (GB)": 1.46,
|
| 38 |
"step": 1500,
|
| 39 |
"total_memory_available (GB)": 126.62
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.058823529411764705,
|
| 43 |
+
"grad_norm": 5.18107271194458,
|
| 44 |
"learning_rate": 4.901960784313725e-05,
|
| 45 |
+
"loss": 1.4003,
|
| 46 |
+
"max_memory_allocated (GB)": 2.1,
|
| 47 |
+
"memory_allocated (GB)": 1.46,
|
| 48 |
"step": 2000,
|
| 49 |
"total_memory_available (GB)": 126.62
|
| 50 |
},
|
| 51 |
{
|
| 52 |
"epoch": 0.07352941176470588,
|
| 53 |
+
"grad_norm": 4.011467933654785,
|
| 54 |
"learning_rate": 4.877450980392157e-05,
|
| 55 |
+
"loss": 1.3578,
|
| 56 |
+
"max_memory_allocated (GB)": 2.1,
|
| 57 |
+
"memory_allocated (GB)": 1.46,
|
| 58 |
"step": 2500,
|
| 59 |
"total_memory_available (GB)": 126.62
|
| 60 |
},
|
| 61 |
{
|
| 62 |
"epoch": 0.08823529411764706,
|
| 63 |
+
"grad_norm": 3.5927531719207764,
|
| 64 |
"learning_rate": 4.8529411764705885e-05,
|
| 65 |
+
"loss": 1.2995,
|
| 66 |
+
"max_memory_allocated (GB)": 2.1,
|
| 67 |
+
"memory_allocated (GB)": 1.46,
|
| 68 |
"step": 3000,
|
| 69 |
"total_memory_available (GB)": 126.62
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.10294117647058823,
|
| 73 |
+
"grad_norm": 4.120026111602783,
|
| 74 |
"learning_rate": 4.82843137254902e-05,
|
| 75 |
+
"loss": 1.2598,
|
| 76 |
+
"max_memory_allocated (GB)": 2.1,
|
| 77 |
+
"memory_allocated (GB)": 1.46,
|
| 78 |
"step": 3500,
|
| 79 |
"total_memory_available (GB)": 126.62
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.11764705882352941,
|
| 83 |
+
"grad_norm": 6.63548469543457,
|
| 84 |
"learning_rate": 4.803921568627452e-05,
|
| 85 |
+
"loss": 1.2075,
|
| 86 |
+
"max_memory_allocated (GB)": 2.1,
|
| 87 |
+
"memory_allocated (GB)": 1.46,
|
| 88 |
"step": 4000,
|
| 89 |
"total_memory_available (GB)": 126.62
|
| 90 |
},
|
| 91 |
{
|
| 92 |
"epoch": 0.1323529411764706,
|
| 93 |
+
"grad_norm": 4.13482666015625,
|
| 94 |
"learning_rate": 4.7794117647058826e-05,
|
| 95 |
+
"loss": 1.2405,
|
| 96 |
+
"max_memory_allocated (GB)": 2.1,
|
| 97 |
+
"memory_allocated (GB)": 1.46,
|
| 98 |
"step": 4500,
|
| 99 |
"total_memory_available (GB)": 126.62
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.14705882352941177,
|
| 103 |
+
"grad_norm": 5.699426174163818,
|
| 104 |
"learning_rate": 4.7549019607843135e-05,
|
| 105 |
+
"loss": 1.2242,
|
| 106 |
+
"max_memory_allocated (GB)": 2.1,
|
| 107 |
+
"memory_allocated (GB)": 1.46,
|
| 108 |
"step": 5000,
|
| 109 |
"total_memory_available (GB)": 126.62
|
| 110 |
},
|
| 111 |
{
|
| 112 |
"epoch": 0.16176470588235295,
|
| 113 |
+
"grad_norm": 6.220912933349609,
|
| 114 |
"learning_rate": 4.730392156862745e-05,
|
| 115 |
+
"loss": 1.1701,
|
| 116 |
+
"max_memory_allocated (GB)": 2.1,
|
| 117 |
+
"memory_allocated (GB)": 1.46,
|
| 118 |
"step": 5500,
|
| 119 |
"total_memory_available (GB)": 126.62
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 0.17647058823529413,
|
| 123 |
+
"grad_norm": 5.078765869140625,
|
| 124 |
"learning_rate": 4.705882352941177e-05,
|
| 125 |
+
"loss": 1.1904,
|
| 126 |
+
"max_memory_allocated (GB)": 2.1,
|
| 127 |
+
"memory_allocated (GB)": 1.46,
|
| 128 |
"step": 6000,
|
| 129 |
"total_memory_available (GB)": 126.62
|
| 130 |
},
|
| 131 |
{
|
| 132 |
"epoch": 0.19117647058823528,
|
| 133 |
+
"grad_norm": 1.5997480154037476,
|
| 134 |
"learning_rate": 4.681372549019608e-05,
|
| 135 |
+
"loss": 1.1321,
|
| 136 |
+
"max_memory_allocated (GB)": 2.1,
|
| 137 |
+
"memory_allocated (GB)": 1.46,
|
| 138 |
"step": 6500,
|
| 139 |
"total_memory_available (GB)": 126.62
|
| 140 |
},
|
| 141 |
{
|
| 142 |
"epoch": 0.20588235294117646,
|
| 143 |
+
"grad_norm": 3.2184040546417236,
|
| 144 |
"learning_rate": 4.656862745098039e-05,
|
| 145 |
+
"loss": 1.1452,
|
| 146 |
+
"max_memory_allocated (GB)": 2.1,
|
| 147 |
+
"memory_allocated (GB)": 1.46,
|
| 148 |
"step": 7000,
|
| 149 |
"total_memory_available (GB)": 126.62
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 0.22058823529411764,
|
| 153 |
+
"grad_norm": 5.408567905426025,
|
| 154 |
"learning_rate": 4.632352941176471e-05,
|
| 155 |
+
"loss": 1.0764,
|
| 156 |
+
"max_memory_allocated (GB)": 2.1,
|
| 157 |
+
"memory_allocated (GB)": 1.46,
|
| 158 |
"step": 7500,
|
| 159 |
"total_memory_available (GB)": 126.62
|
| 160 |
},
|
| 161 |
{
|
| 162 |
"epoch": 0.23529411764705882,
|
| 163 |
+
"grad_norm": 6.1005539894104,
|
| 164 |
"learning_rate": 4.607843137254902e-05,
|
| 165 |
+
"loss": 1.0756,
|
| 166 |
+
"max_memory_allocated (GB)": 2.1,
|
| 167 |
+
"memory_allocated (GB)": 1.46,
|
| 168 |
"step": 8000,
|
| 169 |
"total_memory_available (GB)": 126.62
|
| 170 |
},
|
| 171 |
{
|
| 172 |
"epoch": 0.25,
|
| 173 |
+
"grad_norm": 10.336429595947266,
|
| 174 |
"learning_rate": 4.5833333333333334e-05,
|
| 175 |
+
"loss": 1.084,
|
| 176 |
+
"max_memory_allocated (GB)": 2.1,
|
| 177 |
+
"memory_allocated (GB)": 1.46,
|
| 178 |
"step": 8500,
|
| 179 |
"total_memory_available (GB)": 126.62
|
| 180 |
},
|
| 181 |
{
|
| 182 |
"epoch": 0.2647058823529412,
|
| 183 |
+
"grad_norm": 5.947171211242676,
|
| 184 |
"learning_rate": 4.558823529411765e-05,
|
| 185 |
+
"loss": 1.095,
|
| 186 |
+
"max_memory_allocated (GB)": 2.1,
|
| 187 |
+
"memory_allocated (GB)": 1.46,
|
| 188 |
"step": 9000,
|
| 189 |
"total_memory_available (GB)": 126.62
|
| 190 |
},
|
| 191 |
{
|
| 192 |
"epoch": 0.27941176470588236,
|
| 193 |
+
"grad_norm": 3.9784250259399414,
|
| 194 |
"learning_rate": 4.5343137254901966e-05,
|
| 195 |
+
"loss": 1.0716,
|
| 196 |
+
"max_memory_allocated (GB)": 2.1,
|
| 197 |
+
"memory_allocated (GB)": 1.46,
|
| 198 |
"step": 9500,
|
| 199 |
"total_memory_available (GB)": 126.62
|
| 200 |
},
|
| 201 |
{
|
| 202 |
"epoch": 0.29411764705882354,
|
| 203 |
+
"grad_norm": 7.9583420753479,
|
| 204 |
"learning_rate": 4.5098039215686275e-05,
|
| 205 |
+
"loss": 1.0725,
|
| 206 |
+
"max_memory_allocated (GB)": 2.1,
|
| 207 |
+
"memory_allocated (GB)": 1.46,
|
| 208 |
"step": 10000,
|
| 209 |
"total_memory_available (GB)": 126.62
|
| 210 |
},
|
| 211 |
{
|
| 212 |
"epoch": 0.3088235294117647,
|
| 213 |
+
"grad_norm": 5.341268062591553,
|
| 214 |
"learning_rate": 4.485294117647059e-05,
|
| 215 |
+
"loss": 1.0656,
|
| 216 |
+
"max_memory_allocated (GB)": 2.1,
|
| 217 |
+
"memory_allocated (GB)": 1.46,
|
| 218 |
"step": 10500,
|
| 219 |
"total_memory_available (GB)": 126.62
|
| 220 |
},
|
| 221 |
{
|
| 222 |
"epoch": 0.3235294117647059,
|
| 223 |
+
"grad_norm": 8.68230152130127,
|
| 224 |
"learning_rate": 4.460784313725491e-05,
|
| 225 |
+
"loss": 0.9894,
|
| 226 |
+
"max_memory_allocated (GB)": 2.1,
|
| 227 |
+
"memory_allocated (GB)": 1.46,
|
| 228 |
"step": 11000,
|
| 229 |
"total_memory_available (GB)": 126.62
|
| 230 |
},
|
| 231 |
{
|
| 232 |
"epoch": 0.3382352941176471,
|
| 233 |
+
"grad_norm": 3.516399621963501,
|
| 234 |
"learning_rate": 4.4362745098039216e-05,
|
| 235 |
+
"loss": 1.0527,
|
| 236 |
+
"max_memory_allocated (GB)": 2.1,
|
| 237 |
+
"memory_allocated (GB)": 1.46,
|
| 238 |
"step": 11500,
|
| 239 |
"total_memory_available (GB)": 126.62
|
| 240 |
},
|
| 241 |
{
|
| 242 |
"epoch": 0.35294117647058826,
|
| 243 |
+
"grad_norm": 5.295083045959473,
|
| 244 |
"learning_rate": 4.411764705882353e-05,
|
| 245 |
+
"loss": 1.0073,
|
| 246 |
+
"max_memory_allocated (GB)": 2.1,
|
| 247 |
+
"memory_allocated (GB)": 1.46,
|
| 248 |
"step": 12000,
|
| 249 |
"total_memory_available (GB)": 126.62
|
| 250 |
},
|
| 251 |
{
|
| 252 |
"epoch": 0.36764705882352944,
|
| 253 |
+
"grad_norm": 3.7748844623565674,
|
| 254 |
"learning_rate": 4.387254901960784e-05,
|
| 255 |
+
"loss": 1.0035,
|
| 256 |
+
"max_memory_allocated (GB)": 2.1,
|
| 257 |
+
"memory_allocated (GB)": 1.46,
|
| 258 |
"step": 12500,
|
| 259 |
"total_memory_available (GB)": 126.62
|
| 260 |
},
|
| 261 |
{
|
| 262 |
"epoch": 0.38235294117647056,
|
| 263 |
+
"grad_norm": 7.610727310180664,
|
| 264 |
"learning_rate": 4.362745098039216e-05,
|
| 265 |
+
"loss": 0.9888,
|
| 266 |
+
"max_memory_allocated (GB)": 2.1,
|
| 267 |
+
"memory_allocated (GB)": 1.46,
|
| 268 |
"step": 13000,
|
| 269 |
"total_memory_available (GB)": 126.62
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 0.39705882352941174,
|
| 273 |
+
"grad_norm": 5.661556243896484,
|
| 274 |
"learning_rate": 4.3382352941176474e-05,
|
| 275 |
+
"loss": 1.005,
|
| 276 |
+
"max_memory_allocated (GB)": 2.1,
|
| 277 |
+
"memory_allocated (GB)": 1.46,
|
| 278 |
"step": 13500,
|
| 279 |
"total_memory_available (GB)": 126.62
|
| 280 |
},
|
| 281 |
{
|
| 282 |
"epoch": 0.4117647058823529,
|
| 283 |
+
"grad_norm": 7.081089496612549,
|
| 284 |
"learning_rate": 4.313725490196079e-05,
|
| 285 |
+
"loss": 0.9848,
|
| 286 |
+
"max_memory_allocated (GB)": 2.1,
|
| 287 |
+
"memory_allocated (GB)": 1.46,
|
| 288 |
"step": 14000,
|
| 289 |
"total_memory_available (GB)": 126.62
|
| 290 |
},
|
| 291 |
{
|
| 292 |
"epoch": 0.4264705882352941,
|
| 293 |
+
"grad_norm": 3.7689192295074463,
|
| 294 |
"learning_rate": 4.28921568627451e-05,
|
| 295 |
+
"loss": 0.9745,
|
| 296 |
+
"max_memory_allocated (GB)": 2.1,
|
| 297 |
+
"memory_allocated (GB)": 1.46,
|
| 298 |
"step": 14500,
|
| 299 |
"total_memory_available (GB)": 126.62
|
| 300 |
},
|
| 301 |
{
|
| 302 |
"epoch": 0.4411764705882353,
|
| 303 |
+
"grad_norm": 3.3003389835357666,
|
| 304 |
"learning_rate": 4.2647058823529415e-05,
|
| 305 |
+
"loss": 0.9684,
|
| 306 |
+
"max_memory_allocated (GB)": 2.1,
|
| 307 |
+
"memory_allocated (GB)": 1.46,
|
| 308 |
"step": 15000,
|
| 309 |
"total_memory_available (GB)": 126.62
|
| 310 |
},
|
| 311 |
{
|
| 312 |
"epoch": 0.45588235294117646,
|
| 313 |
+
"grad_norm": 5.606069564819336,
|
| 314 |
"learning_rate": 4.2401960784313724e-05,
|
| 315 |
+
"loss": 0.9821,
|
| 316 |
+
"max_memory_allocated (GB)": 2.1,
|
| 317 |
+
"memory_allocated (GB)": 1.46,
|
| 318 |
"step": 15500,
|
| 319 |
"total_memory_available (GB)": 126.62
|
| 320 |
},
|
| 321 |
{
|
| 322 |
"epoch": 0.47058823529411764,
|
| 323 |
+
"grad_norm": 5.626067638397217,
|
| 324 |
"learning_rate": 4.215686274509804e-05,
|
| 325 |
+
"loss": 0.9537,
|
| 326 |
+
"max_memory_allocated (GB)": 2.1,
|
| 327 |
+
"memory_allocated (GB)": 1.46,
|
| 328 |
"step": 16000,
|
| 329 |
"total_memory_available (GB)": 126.62
|
| 330 |
},
|
| 331 |
{
|
| 332 |
"epoch": 0.4852941176470588,
|
| 333 |
+
"grad_norm": 5.848409175872803,
|
| 334 |
"learning_rate": 4.1911764705882356e-05,
|
| 335 |
+
"loss": 0.9675,
|
| 336 |
+
"max_memory_allocated (GB)": 2.1,
|
| 337 |
+
"memory_allocated (GB)": 1.46,
|
| 338 |
"step": 16500,
|
| 339 |
"total_memory_available (GB)": 126.62
|
| 340 |
},
|
| 341 |
{
|
| 342 |
"epoch": 0.5,
|
| 343 |
+
"grad_norm": 5.057644844055176,
|
| 344 |
"learning_rate": 4.166666666666667e-05,
|
| 345 |
+
"loss": 0.9714,
|
| 346 |
+
"max_memory_allocated (GB)": 2.1,
|
| 347 |
+
"memory_allocated (GB)": 1.46,
|
| 348 |
"step": 17000,
|
| 349 |
"total_memory_available (GB)": 126.62
|
| 350 |
},
|
| 351 |
{
|
| 352 |
"epoch": 0.5147058823529411,
|
| 353 |
+
"grad_norm": 3.4304769039154053,
|
| 354 |
"learning_rate": 4.142156862745099e-05,
|
| 355 |
+
"loss": 0.9857,
|
| 356 |
+
"max_memory_allocated (GB)": 2.1,
|
| 357 |
+
"memory_allocated (GB)": 1.46,
|
| 358 |
"step": 17500,
|
| 359 |
"total_memory_available (GB)": 126.62
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"epoch": 0.5294117647058824,
|
| 363 |
+
"grad_norm": 3.0925090312957764,
|
| 364 |
"learning_rate": 4.11764705882353e-05,
|
| 365 |
+
"loss": 0.9716,
|
| 366 |
+
"max_memory_allocated (GB)": 2.1,
|
| 367 |
+
"memory_allocated (GB)": 1.46,
|
| 368 |
"step": 18000,
|
| 369 |
"total_memory_available (GB)": 126.62
|
| 370 |
},
|
| 371 |
{
|
| 372 |
"epoch": 0.5441176470588235,
|
| 373 |
+
"grad_norm": 7.13212776184082,
|
| 374 |
"learning_rate": 4.0931372549019607e-05,
|
| 375 |
+
"loss": 0.9471,
|
| 376 |
+
"max_memory_allocated (GB)": 2.1,
|
| 377 |
+
"memory_allocated (GB)": 1.46,
|
| 378 |
"step": 18500,
|
| 379 |
"total_memory_available (GB)": 126.62
|
| 380 |
},
|
| 381 |
{
|
| 382 |
"epoch": 0.5588235294117647,
|
| 383 |
+
"grad_norm": 3.437065362930298,
|
| 384 |
"learning_rate": 4.068627450980392e-05,
|
| 385 |
+
"loss": 0.9125,
|
| 386 |
+
"max_memory_allocated (GB)": 2.1,
|
| 387 |
+
"memory_allocated (GB)": 1.46,
|
| 388 |
"step": 19000,
|
| 389 |
"total_memory_available (GB)": 126.62
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 0.5735294117647058,
|
| 393 |
+
"grad_norm": 3.033743143081665,
|
| 394 |
"learning_rate": 4.044117647058824e-05,
|
| 395 |
+
"loss": 0.9608,
|
| 396 |
+
"max_memory_allocated (GB)": 2.1,
|
| 397 |
+
"memory_allocated (GB)": 1.46,
|
| 398 |
"step": 19500,
|
| 399 |
"total_memory_available (GB)": 126.62
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 0.5882352941176471,
|
| 403 |
+
"grad_norm": 4.206902980804443,
|
| 404 |
"learning_rate": 4.0196078431372555e-05,
|
| 405 |
+
"loss": 0.9494,
|
| 406 |
+
"max_memory_allocated (GB)": 2.1,
|
| 407 |
+
"memory_allocated (GB)": 1.46,
|
| 408 |
"step": 20000,
|
| 409 |
"total_memory_available (GB)": 126.62
|
| 410 |
},
|
| 411 |
{
|
| 412 |
"epoch": 0.6029411764705882,
|
| 413 |
+
"grad_norm": 4.862374305725098,
|
| 414 |
"learning_rate": 3.9950980392156864e-05,
|
| 415 |
+
"loss": 0.9636,
|
| 416 |
+
"max_memory_allocated (GB)": 2.1,
|
| 417 |
+
"memory_allocated (GB)": 1.46,
|
| 418 |
"step": 20500,
|
| 419 |
"total_memory_available (GB)": 126.62
|
| 420 |
},
|
| 421 |
{
|
| 422 |
"epoch": 0.6176470588235294,
|
| 423 |
+
"grad_norm": 2.435167074203491,
|
| 424 |
"learning_rate": 3.970588235294117e-05,
|
| 425 |
+
"loss": 0.947,
|
| 426 |
+
"max_memory_allocated (GB)": 2.1,
|
| 427 |
+
"memory_allocated (GB)": 1.46,
|
| 428 |
"step": 21000,
|
| 429 |
"total_memory_available (GB)": 126.62
|
| 430 |
},
|
| 431 |
{
|
| 432 |
"epoch": 0.6323529411764706,
|
| 433 |
+
"grad_norm": 3.3232932090759277,
|
| 434 |
"learning_rate": 3.946078431372549e-05,
|
| 435 |
+
"loss": 0.9104,
|
| 436 |
+
"max_memory_allocated (GB)": 2.1,
|
| 437 |
+
"memory_allocated (GB)": 1.46,
|
| 438 |
"step": 21500,
|
| 439 |
"total_memory_available (GB)": 126.62
|
| 440 |
},
|
| 441 |
{
|
| 442 |
"epoch": 0.6470588235294118,
|
| 443 |
+
"grad_norm": 6.478214740753174,
|
| 444 |
"learning_rate": 3.9215686274509805e-05,
|
| 445 |
+
"loss": 0.9457,
|
| 446 |
+
"max_memory_allocated (GB)": 2.1,
|
| 447 |
+
"memory_allocated (GB)": 1.46,
|
| 448 |
"step": 22000,
|
| 449 |
"total_memory_available (GB)": 126.62
|
| 450 |
},
|
| 451 |
{
|
| 452 |
"epoch": 0.6617647058823529,
|
| 453 |
+
"grad_norm": 4.218869209289551,
|
| 454 |
"learning_rate": 3.897058823529412e-05,
|
| 455 |
+
"loss": 0.9154,
|
| 456 |
+
"max_memory_allocated (GB)": 2.1,
|
| 457 |
+
"memory_allocated (GB)": 1.46,
|
| 458 |
"step": 22500,
|
| 459 |
"total_memory_available (GB)": 126.62
|
| 460 |
},
|
| 461 |
{
|
| 462 |
"epoch": 0.6764705882352942,
|
| 463 |
+
"grad_norm": 3.7748098373413086,
|
| 464 |
"learning_rate": 3.872549019607844e-05,
|
| 465 |
+
"loss": 0.9,
|
| 466 |
+
"max_memory_allocated (GB)": 2.1,
|
| 467 |
+
"memory_allocated (GB)": 1.46,
|
| 468 |
"step": 23000,
|
| 469 |
"total_memory_available (GB)": 126.62
|
| 470 |
},
|
| 471 |
{
|
| 472 |
"epoch": 0.6911764705882353,
|
| 473 |
+
"grad_norm": 4.164327621459961,
|
| 474 |
"learning_rate": 3.8480392156862746e-05,
|
| 475 |
+
"loss": 0.9025,
|
| 476 |
+
"max_memory_allocated (GB)": 2.1,
|
| 477 |
+
"memory_allocated (GB)": 1.46,
|
| 478 |
"step": 23500,
|
| 479 |
"total_memory_available (GB)": 126.62
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 0.7058823529411765,
|
| 483 |
+
"grad_norm": 4.087977886199951,
|
| 484 |
"learning_rate": 3.8235294117647055e-05,
|
| 485 |
+
"loss": 0.9147,
|
| 486 |
+
"max_memory_allocated (GB)": 2.1,
|
| 487 |
+
"memory_allocated (GB)": 1.46,
|
| 488 |
"step": 24000,
|
| 489 |
"total_memory_available (GB)": 126.62
|
| 490 |
},
|
| 491 |
{
|
| 492 |
"epoch": 0.7205882352941176,
|
| 493 |
+
"grad_norm": 7.114366054534912,
|
| 494 |
"learning_rate": 3.799019607843137e-05,
|
| 495 |
+
"loss": 0.8656,
|
| 496 |
+
"max_memory_allocated (GB)": 2.1,
|
| 497 |
+
"memory_allocated (GB)": 1.46,
|
| 498 |
"step": 24500,
|
| 499 |
"total_memory_available (GB)": 126.62
|
| 500 |
},
|
| 501 |
{
|
| 502 |
"epoch": 0.7352941176470589,
|
| 503 |
+
"grad_norm": 6.7599101066589355,
|
| 504 |
"learning_rate": 3.774509803921569e-05,
|
| 505 |
+
"loss": 0.8887,
|
| 506 |
+
"max_memory_allocated (GB)": 2.1,
|
| 507 |
+
"memory_allocated (GB)": 1.46,
|
| 508 |
"step": 25000,
|
| 509 |
"total_memory_available (GB)": 126.62
|
| 510 |
},
|
| 511 |
{
|
| 512 |
"epoch": 0.75,
|
| 513 |
+
"grad_norm": 5.847716808319092,
|
| 514 |
"learning_rate": 3.7500000000000003e-05,
|
| 515 |
+
"loss": 0.8807,
|
| 516 |
+
"max_memory_allocated (GB)": 2.1,
|
| 517 |
+
"memory_allocated (GB)": 1.46,
|
| 518 |
"step": 25500,
|
| 519 |
"total_memory_available (GB)": 126.62
|
| 520 |
},
|
| 521 |
{
|
| 522 |
"epoch": 0.7647058823529411,
|
| 523 |
+
"grad_norm": 9.692086219787598,
|
| 524 |
"learning_rate": 3.725490196078432e-05,
|
| 525 |
+
"loss": 0.8615,
|
| 526 |
+
"max_memory_allocated (GB)": 2.1,
|
| 527 |
+
"memory_allocated (GB)": 1.46,
|
| 528 |
"step": 26000,
|
| 529 |
"total_memory_available (GB)": 126.62
|
| 530 |
},
|
| 531 |
{
|
| 532 |
"epoch": 0.7794117647058824,
|
| 533 |
+
"grad_norm": 0.6128278970718384,
|
| 534 |
"learning_rate": 3.700980392156863e-05,
|
| 535 |
+
"loss": 0.8675,
|
| 536 |
+
"max_memory_allocated (GB)": 2.1,
|
| 537 |
+
"memory_allocated (GB)": 1.46,
|
| 538 |
"step": 26500,
|
| 539 |
"total_memory_available (GB)": 126.62
|
| 540 |
},
|
| 541 |
{
|
| 542 |
"epoch": 0.7941176470588235,
|
| 543 |
+
"grad_norm": 8.303850173950195,
|
| 544 |
"learning_rate": 3.6764705882352945e-05,
|
| 545 |
+
"loss": 0.8809,
|
| 546 |
+
"max_memory_allocated (GB)": 2.1,
|
| 547 |
+
"memory_allocated (GB)": 1.46,
|
| 548 |
"step": 27000,
|
| 549 |
"total_memory_available (GB)": 126.62
|
| 550 |
},
|
| 551 |
{
|
| 552 |
"epoch": 0.8088235294117647,
|
| 553 |
+
"grad_norm": 3.569289445877075,
|
| 554 |
"learning_rate": 3.6519607843137254e-05,
|
| 555 |
+
"loss": 0.8891,
|
| 556 |
+
"max_memory_allocated (GB)": 2.1,
|
| 557 |
+
"memory_allocated (GB)": 1.46,
|
| 558 |
"step": 27500,
|
| 559 |
"total_memory_available (GB)": 126.62
|
| 560 |
},
|
| 561 |
{
|
| 562 |
"epoch": 0.8235294117647058,
|
| 563 |
+
"grad_norm": 0.8430338501930237,
|
| 564 |
"learning_rate": 3.627450980392157e-05,
|
| 565 |
+
"loss": 0.8613,
|
| 566 |
+
"max_memory_allocated (GB)": 2.1,
|
| 567 |
+
"memory_allocated (GB)": 1.46,
|
| 568 |
"step": 28000,
|
| 569 |
"total_memory_available (GB)": 126.62
|
| 570 |
},
|
| 571 |
{
|
| 572 |
"epoch": 0.8382352941176471,
|
| 573 |
+
"grad_norm": 3.0527398586273193,
|
| 574 |
"learning_rate": 3.6029411764705886e-05,
|
| 575 |
+
"loss": 0.8622,
|
| 576 |
+
"max_memory_allocated (GB)": 2.1,
|
| 577 |
+
"memory_allocated (GB)": 1.46,
|
| 578 |
"step": 28500,
|
| 579 |
"total_memory_available (GB)": 126.62
|
| 580 |
},
|
| 581 |
{
|
| 582 |
"epoch": 0.8529411764705882,
|
| 583 |
+
"grad_norm": 7.135374069213867,
|
| 584 |
"learning_rate": 3.5784313725490195e-05,
|
| 585 |
+
"loss": 0.8671,
|
| 586 |
+
"max_memory_allocated (GB)": 2.1,
|
| 587 |
+
"memory_allocated (GB)": 1.46,
|
| 588 |
"step": 29000,
|
| 589 |
"total_memory_available (GB)": 126.62
|
| 590 |
},
|
| 591 |
{
|
| 592 |
"epoch": 0.8676470588235294,
|
| 593 |
+
"grad_norm": 3.676905393600464,
|
| 594 |
"learning_rate": 3.553921568627451e-05,
|
| 595 |
+
"loss": 0.8788,
|
| 596 |
+
"max_memory_allocated (GB)": 2.1,
|
| 597 |
+
"memory_allocated (GB)": 1.46,
|
| 598 |
"step": 29500,
|
| 599 |
"total_memory_available (GB)": 126.62
|
| 600 |
},
|
| 601 |
{
|
| 602 |
"epoch": 0.8823529411764706,
|
| 603 |
+
"grad_norm": 7.306654453277588,
|
| 604 |
"learning_rate": 3.529411764705883e-05,
|
| 605 |
+
"loss": 0.8283,
|
| 606 |
+
"max_memory_allocated (GB)": 2.1,
|
| 607 |
+
"memory_allocated (GB)": 1.46,
|
| 608 |
"step": 30000,
|
| 609 |
"total_memory_available (GB)": 126.62
|
| 610 |
},
|
| 611 |
{
|
| 612 |
"epoch": 0.8970588235294118,
|
| 613 |
+
"grad_norm": 6.978466510772705,
|
| 614 |
"learning_rate": 3.5049019607843136e-05,
|
| 615 |
+
"loss": 0.8755,
|
| 616 |
+
"max_memory_allocated (GB)": 2.1,
|
| 617 |
+
"memory_allocated (GB)": 1.46,
|
| 618 |
"step": 30500,
|
| 619 |
"total_memory_available (GB)": 126.62
|
| 620 |
},
|
| 621 |
{
|
| 622 |
"epoch": 0.9117647058823529,
|
| 623 |
+
"grad_norm": 8.922977447509766,
|
| 624 |
"learning_rate": 3.480392156862745e-05,
|
| 625 |
+
"loss": 0.8398,
|
| 626 |
+
"max_memory_allocated (GB)": 2.1,
|
| 627 |
+
"memory_allocated (GB)": 1.46,
|
| 628 |
"step": 31000,
|
| 629 |
"total_memory_available (GB)": 126.62
|
| 630 |
},
|
| 631 |
{
|
| 632 |
"epoch": 0.9264705882352942,
|
| 633 |
+
"grad_norm": 3.4424359798431396,
|
| 634 |
"learning_rate": 3.455882352941177e-05,
|
| 635 |
+
"loss": 0.8565,
|
| 636 |
+
"max_memory_allocated (GB)": 2.1,
|
| 637 |
+
"memory_allocated (GB)": 1.46,
|
| 638 |
"step": 31500,
|
| 639 |
"total_memory_available (GB)": 126.62
|
| 640 |
},
|
| 641 |
{
|
| 642 |
"epoch": 0.9411764705882353,
|
| 643 |
+
"grad_norm": 4.229028701782227,
|
| 644 |
"learning_rate": 3.431372549019608e-05,
|
| 645 |
+
"loss": 0.851,
|
| 646 |
+
"max_memory_allocated (GB)": 2.1,
|
| 647 |
+
"memory_allocated (GB)": 1.46,
|
| 648 |
"step": 32000,
|
| 649 |
"total_memory_available (GB)": 126.62
|
| 650 |
},
|
| 651 |
{
|
| 652 |
"epoch": 0.9558823529411765,
|
| 653 |
+
"grad_norm": 12.309649467468262,
|
| 654 |
"learning_rate": 3.4068627450980394e-05,
|
| 655 |
+
"loss": 0.8563,
|
| 656 |
+
"max_memory_allocated (GB)": 2.1,
|
| 657 |
+
"memory_allocated (GB)": 1.46,
|
| 658 |
"step": 32500,
|
| 659 |
"total_memory_available (GB)": 126.62
|
| 660 |
},
|
| 661 |
{
|
| 662 |
"epoch": 0.9705882352941176,
|
| 663 |
+
"grad_norm": 0.5380654335021973,
|
| 664 |
"learning_rate": 3.382352941176471e-05,
|
| 665 |
+
"loss": 0.8338,
|
| 666 |
+
"max_memory_allocated (GB)": 2.1,
|
| 667 |
+
"memory_allocated (GB)": 1.46,
|
| 668 |
"step": 33000,
|
| 669 |
"total_memory_available (GB)": 126.62
|
| 670 |
},
|
| 671 |
{
|
| 672 |
"epoch": 0.9852941176470589,
|
| 673 |
+
"grad_norm": 4.774703025817871,
|
| 674 |
"learning_rate": 3.357843137254902e-05,
|
| 675 |
+
"loss": 0.8425,
|
| 676 |
+
"max_memory_allocated (GB)": 2.1,
|
| 677 |
+
"memory_allocated (GB)": 1.46,
|
| 678 |
"step": 33500,
|
| 679 |
"total_memory_available (GB)": 126.62
|
| 680 |
},
|
| 681 |
{
|
| 682 |
"epoch": 1.0,
|
| 683 |
+
"grad_norm": 5.663310527801514,
|
| 684 |
"learning_rate": 3.3333333333333335e-05,
|
| 685 |
+
"loss": 0.8711,
|
| 686 |
+
"max_memory_allocated (GB)": 2.1,
|
| 687 |
+
"memory_allocated (GB)": 1.46,
|
| 688 |
"step": 34000,
|
| 689 |
"total_memory_available (GB)": 126.62
|
| 690 |
},
|
| 691 |
{
|
| 692 |
"epoch": 1.0147058823529411,
|
| 693 |
+
"grad_norm": 5.9944562911987305,
|
| 694 |
"learning_rate": 3.308823529411765e-05,
|
| 695 |
+
"loss": 0.8495,
|
| 696 |
+
"max_memory_allocated (GB)": 2.1,
|
| 697 |
+
"memory_allocated (GB)": 1.46,
|
| 698 |
"step": 34500,
|
| 699 |
"total_memory_available (GB)": 126.62
|
| 700 |
},
|
| 701 |
{
|
| 702 |
"epoch": 1.0294117647058822,
|
| 703 |
+
"grad_norm": 4.68066930770874,
|
| 704 |
"learning_rate": 3.284313725490196e-05,
|
| 705 |
+
"loss": 0.7897,
|
| 706 |
+
"max_memory_allocated (GB)": 2.1,
|
| 707 |
+
"memory_allocated (GB)": 1.46,
|
| 708 |
"step": 35000,
|
| 709 |
"total_memory_available (GB)": 126.62
|
| 710 |
},
|
| 711 |
{
|
| 712 |
"epoch": 1.0441176470588236,
|
| 713 |
+
"grad_norm": 1.811689853668213,
|
| 714 |
"learning_rate": 3.2598039215686276e-05,
|
| 715 |
+
"loss": 0.8056,
|
| 716 |
+
"max_memory_allocated (GB)": 2.1,
|
| 717 |
+
"memory_allocated (GB)": 1.46,
|
| 718 |
"step": 35500,
|
| 719 |
"total_memory_available (GB)": 126.62
|
| 720 |
},
|
| 721 |
{
|
| 722 |
"epoch": 1.0588235294117647,
|
| 723 |
+
"grad_norm": 5.509186744689941,
|
| 724 |
"learning_rate": 3.235294117647059e-05,
|
| 725 |
+
"loss": 0.8267,
|
| 726 |
+
"max_memory_allocated (GB)": 2.1,
|
| 727 |
+
"memory_allocated (GB)": 1.46,
|
| 728 |
"step": 36000,
|
| 729 |
"total_memory_available (GB)": 126.62
|
| 730 |
},
|
| 731 |
{
|
| 732 |
"epoch": 1.0735294117647058,
|
| 733 |
+
"grad_norm": 9.382353782653809,
|
| 734 |
"learning_rate": 3.210784313725491e-05,
|
| 735 |
+
"loss": 0.8298,
|
| 736 |
+
"max_memory_allocated (GB)": 2.1,
|
| 737 |
+
"memory_allocated (GB)": 1.46,
|
| 738 |
"step": 36500,
|
| 739 |
"total_memory_available (GB)": 126.62
|
| 740 |
},
|
| 741 |
{
|
| 742 |
"epoch": 1.088235294117647,
|
| 743 |
+
"grad_norm": 5.186063766479492,
|
| 744 |
"learning_rate": 3.186274509803922e-05,
|
| 745 |
+
"loss": 0.8466,
|
| 746 |
+
"max_memory_allocated (GB)": 2.1,
|
| 747 |
+
"memory_allocated (GB)": 1.46,
|
| 748 |
"step": 37000,
|
| 749 |
"total_memory_available (GB)": 126.62
|
| 750 |
},
|
| 751 |
{
|
| 752 |
"epoch": 1.1029411764705883,
|
| 753 |
+
"grad_norm": 6.467650413513184,
|
| 754 |
"learning_rate": 3.161764705882353e-05,
|
| 755 |
+
"loss": 0.8027,
|
| 756 |
+
"max_memory_allocated (GB)": 2.1,
|
| 757 |
+
"memory_allocated (GB)": 1.46,
|
| 758 |
"step": 37500,
|
| 759 |
"total_memory_available (GB)": 126.62
|
| 760 |
},
|
| 761 |
{
|
| 762 |
"epoch": 1.1176470588235294,
|
| 763 |
+
"grad_norm": 4.107296943664551,
|
| 764 |
"learning_rate": 3.137254901960784e-05,
|
| 765 |
+
"loss": 0.7936,
|
| 766 |
+
"max_memory_allocated (GB)": 2.1,
|
| 767 |
+
"memory_allocated (GB)": 1.46,
|
| 768 |
"step": 38000,
|
| 769 |
"total_memory_available (GB)": 126.62
|
| 770 |
},
|
| 771 |
{
|
| 772 |
"epoch": 1.1323529411764706,
|
| 773 |
+
"grad_norm": 2.7116880416870117,
|
| 774 |
"learning_rate": 3.112745098039216e-05,
|
| 775 |
+
"loss": 0.7658,
|
| 776 |
+
"max_memory_allocated (GB)": 2.1,
|
| 777 |
+
"memory_allocated (GB)": 1.46,
|
| 778 |
"step": 38500,
|
| 779 |
"total_memory_available (GB)": 126.62
|
| 780 |
},
|
| 781 |
{
|
| 782 |
"epoch": 1.1470588235294117,
|
| 783 |
+
"grad_norm": 4.101169109344482,
|
| 784 |
"learning_rate": 3.0882352941176475e-05,
|
| 785 |
+
"loss": 0.8117,
|
| 786 |
+
"max_memory_allocated (GB)": 2.1,
|
| 787 |
+
"memory_allocated (GB)": 1.46,
|
| 788 |
"step": 39000,
|
| 789 |
"total_memory_available (GB)": 126.62
|
| 790 |
},
|
| 791 |
{
|
| 792 |
"epoch": 1.161764705882353,
|
| 793 |
+
"grad_norm": 4.81936502456665,
|
| 794 |
"learning_rate": 3.063725490196079e-05,
|
| 795 |
+
"loss": 0.8001,
|
| 796 |
+
"max_memory_allocated (GB)": 2.1,
|
| 797 |
+
"memory_allocated (GB)": 1.46,
|
| 798 |
"step": 39500,
|
| 799 |
"total_memory_available (GB)": 126.62
|
| 800 |
},
|
| 801 |
{
|
| 802 |
"epoch": 1.1764705882352942,
|
| 803 |
+
"grad_norm": 7.343831539154053,
|
| 804 |
"learning_rate": 3.0392156862745097e-05,
|
| 805 |
+
"loss": 0.7704,
|
| 806 |
+
"max_memory_allocated (GB)": 2.1,
|
| 807 |
+
"memory_allocated (GB)": 1.46,
|
| 808 |
"step": 40000,
|
| 809 |
"total_memory_available (GB)": 126.62
|
| 810 |
},
|
| 811 |
{
|
| 812 |
"epoch": 1.1911764705882353,
|
| 813 |
+
"grad_norm": 8.532153129577637,
|
| 814 |
"learning_rate": 3.0147058823529413e-05,
|
| 815 |
+
"loss": 0.7917,
|
| 816 |
+
"max_memory_allocated (GB)": 2.1,
|
| 817 |
+
"memory_allocated (GB)": 1.46,
|
| 818 |
"step": 40500,
|
| 819 |
"total_memory_available (GB)": 126.62
|
| 820 |
},
|
| 821 |
{
|
| 822 |
"epoch": 1.2058823529411764,
|
| 823 |
+
"grad_norm": 7.05020809173584,
|
| 824 |
"learning_rate": 2.9901960784313725e-05,
|
| 825 |
+
"loss": 0.808,
|
| 826 |
+
"max_memory_allocated (GB)": 2.1,
|
| 827 |
+
"memory_allocated (GB)": 1.46,
|
| 828 |
"step": 41000,
|
| 829 |
"total_memory_available (GB)": 126.62
|
| 830 |
},
|
| 831 |
{
|
| 832 |
"epoch": 1.2205882352941178,
|
| 833 |
+
"grad_norm": 3.989778995513916,
|
| 834 |
"learning_rate": 2.965686274509804e-05,
|
| 835 |
+
"loss": 0.8121,
|
| 836 |
+
"max_memory_allocated (GB)": 2.1,
|
| 837 |
+
"memory_allocated (GB)": 1.46,
|
| 838 |
"step": 41500,
|
| 839 |
"total_memory_available (GB)": 126.62
|
| 840 |
},
|
| 841 |
{
|
| 842 |
"epoch": 1.2352941176470589,
|
| 843 |
+
"grad_norm": 6.2639689445495605,
|
| 844 |
"learning_rate": 2.9411764705882354e-05,
|
| 845 |
+
"loss": 0.7748,
|
| 846 |
+
"max_memory_allocated (GB)": 2.1,
|
| 847 |
+
"memory_allocated (GB)": 1.46,
|
| 848 |
"step": 42000,
|
| 849 |
"total_memory_available (GB)": 126.62
|
| 850 |
},
|
| 851 |
{
|
| 852 |
"epoch": 1.25,
|
| 853 |
+
"grad_norm": 3.3618924617767334,
|
| 854 |
"learning_rate": 2.916666666666667e-05,
|
| 855 |
+
"loss": 0.813,
|
| 856 |
+
"max_memory_allocated (GB)": 2.1,
|
| 857 |
+
"memory_allocated (GB)": 1.46,
|
| 858 |
"step": 42500,
|
| 859 |
"total_memory_available (GB)": 126.62
|
| 860 |
},
|
| 861 |
{
|
| 862 |
"epoch": 1.2647058823529411,
|
| 863 |
+
"grad_norm": 4.954437255859375,
|
| 864 |
"learning_rate": 2.8921568627450986e-05,
|
| 865 |
+
"loss": 0.7693,
|
| 866 |
+
"max_memory_allocated (GB)": 2.1,
|
| 867 |
+
"memory_allocated (GB)": 1.46,
|
| 868 |
"step": 43000,
|
| 869 |
"total_memory_available (GB)": 126.62
|
| 870 |
},
|
| 871 |
{
|
| 872 |
"epoch": 1.2794117647058822,
|
| 873 |
+
"grad_norm": 3.4438722133636475,
|
| 874 |
"learning_rate": 2.8676470588235295e-05,
|
| 875 |
+
"loss": 0.7994,
|
| 876 |
+
"max_memory_allocated (GB)": 2.1,
|
| 877 |
+
"memory_allocated (GB)": 1.46,
|
| 878 |
"step": 43500,
|
| 879 |
"total_memory_available (GB)": 126.62
|
| 880 |
},
|
| 881 |
{
|
| 882 |
"epoch": 1.2941176470588236,
|
| 883 |
+
"grad_norm": 4.17124080657959,
|
| 884 |
"learning_rate": 2.8431372549019608e-05,
|
| 885 |
+
"loss": 0.7764,
|
| 886 |
+
"max_memory_allocated (GB)": 2.1,
|
| 887 |
+
"memory_allocated (GB)": 1.46,
|
| 888 |
"step": 44000,
|
| 889 |
"total_memory_available (GB)": 126.62
|
| 890 |
},
|
| 891 |
{
|
| 892 |
"epoch": 1.3088235294117647,
|
| 893 |
+
"grad_norm": 8.348847389221191,
|
| 894 |
"learning_rate": 2.8186274509803924e-05,
|
| 895 |
+
"loss": 0.7478,
|
| 896 |
+
"max_memory_allocated (GB)": 2.1,
|
| 897 |
+
"memory_allocated (GB)": 1.46,
|
| 898 |
"step": 44500,
|
| 899 |
"total_memory_available (GB)": 126.62
|
| 900 |
},
|
| 901 |
{
|
| 902 |
"epoch": 1.3235294117647058,
|
| 903 |
+
"grad_norm": 1.2507061958312988,
|
| 904 |
"learning_rate": 2.7941176470588236e-05,
|
| 905 |
+
"loss": 0.7974,
|
| 906 |
+
"max_memory_allocated (GB)": 2.1,
|
| 907 |
+
"memory_allocated (GB)": 1.46,
|
| 908 |
"step": 45000,
|
| 909 |
"total_memory_available (GB)": 126.62
|
| 910 |
},
|
| 911 |
{
|
| 912 |
"epoch": 1.3382352941176472,
|
| 913 |
+
"grad_norm": 13.898884773254395,
|
| 914 |
"learning_rate": 2.7696078431372552e-05,
|
| 915 |
+
"loss": 0.7541,
|
| 916 |
+
"max_memory_allocated (GB)": 2.1,
|
| 917 |
+
"memory_allocated (GB)": 1.46,
|
| 918 |
"step": 45500,
|
| 919 |
"total_memory_available (GB)": 126.62
|
| 920 |
},
|
| 921 |
{
|
| 922 |
"epoch": 1.3529411764705883,
|
| 923 |
+
"grad_norm": 7.0425872802734375,
|
| 924 |
"learning_rate": 2.7450980392156865e-05,
|
| 925 |
+
"loss": 0.7837,
|
| 926 |
+
"max_memory_allocated (GB)": 2.1,
|
| 927 |
+
"memory_allocated (GB)": 1.46,
|
| 928 |
"step": 46000,
|
| 929 |
"total_memory_available (GB)": 126.62
|
| 930 |
},
|
| 931 |
{
|
| 932 |
"epoch": 1.3676470588235294,
|
| 933 |
+
"grad_norm": 5.521500587463379,
|
| 934 |
"learning_rate": 2.7205882352941174e-05,
|
| 935 |
+
"loss": 0.7847,
|
| 936 |
+
"max_memory_allocated (GB)": 2.1,
|
| 937 |
+
"memory_allocated (GB)": 1.46,
|
| 938 |
"step": 46500,
|
| 939 |
"total_memory_available (GB)": 126.62
|
| 940 |
},
|
| 941 |
{
|
| 942 |
"epoch": 1.3823529411764706,
|
| 943 |
+
"grad_norm": 7.491474628448486,
|
| 944 |
"learning_rate": 2.696078431372549e-05,
|
| 945 |
+
"loss": 0.748,
|
| 946 |
+
"max_memory_allocated (GB)": 2.1,
|
| 947 |
+
"memory_allocated (GB)": 1.46,
|
| 948 |
"step": 47000,
|
| 949 |
"total_memory_available (GB)": 126.62
|
| 950 |
},
|
| 951 |
{
|
| 952 |
"epoch": 1.3970588235294117,
|
| 953 |
+
"grad_norm": 10.513397216796875,
|
| 954 |
"learning_rate": 2.6715686274509806e-05,
|
| 955 |
+
"loss": 0.7801,
|
| 956 |
+
"max_memory_allocated (GB)": 2.1,
|
| 957 |
+
"memory_allocated (GB)": 1.46,
|
| 958 |
"step": 47500,
|
| 959 |
"total_memory_available (GB)": 126.62
|
| 960 |
},
|
| 961 |
{
|
| 962 |
"epoch": 1.4117647058823528,
|
| 963 |
+
"grad_norm": 15.570920944213867,
|
| 964 |
"learning_rate": 2.647058823529412e-05,
|
| 965 |
+
"loss": 0.7697,
|
| 966 |
+
"max_memory_allocated (GB)": 2.1,
|
| 967 |
+
"memory_allocated (GB)": 1.46,
|
| 968 |
"step": 48000,
|
| 969 |
"total_memory_available (GB)": 126.62
|
| 970 |
},
|
| 971 |
{
|
| 972 |
"epoch": 1.4264705882352942,
|
| 973 |
+
"grad_norm": 4.1163010597229,
|
| 974 |
"learning_rate": 2.6225490196078435e-05,
|
| 975 |
+
"loss": 0.749,
|
| 976 |
+
"max_memory_allocated (GB)": 2.1,
|
| 977 |
+
"memory_allocated (GB)": 1.46,
|
| 978 |
"step": 48500,
|
| 979 |
"total_memory_available (GB)": 126.62
|
| 980 |
},
|
| 981 |
{
|
| 982 |
"epoch": 1.4411764705882353,
|
| 983 |
+
"grad_norm": 7.289860248565674,
|
| 984 |
"learning_rate": 2.5980392156862747e-05,
|
| 985 |
+
"loss": 0.7788,
|
| 986 |
+
"max_memory_allocated (GB)": 2.1,
|
| 987 |
+
"memory_allocated (GB)": 1.46,
|
| 988 |
"step": 49000,
|
| 989 |
"total_memory_available (GB)": 126.62
|
| 990 |
},
|
| 991 |
{
|
| 992 |
"epoch": 1.4558823529411764,
|
| 993 |
+
"grad_norm": 5.316174507141113,
|
| 994 |
"learning_rate": 2.5735294117647057e-05,
|
| 995 |
+
"loss": 0.8029,
|
| 996 |
+
"max_memory_allocated (GB)": 2.1,
|
| 997 |
+
"memory_allocated (GB)": 1.46,
|
| 998 |
"step": 49500,
|
| 999 |
"total_memory_available (GB)": 126.62
|
| 1000 |
},
|
| 1001 |
{
|
| 1002 |
"epoch": 1.4705882352941178,
|
| 1003 |
+
"grad_norm": 3.60591721534729,
|
| 1004 |
"learning_rate": 2.5490196078431373e-05,
|
| 1005 |
+
"loss": 0.7372,
|
| 1006 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1007 |
+
"memory_allocated (GB)": 1.46,
|
| 1008 |
"step": 50000,
|
| 1009 |
"total_memory_available (GB)": 126.62
|
| 1010 |
},
|
| 1011 |
{
|
| 1012 |
"epoch": 1.4852941176470589,
|
| 1013 |
+
"grad_norm": 5.769986152648926,
|
| 1014 |
"learning_rate": 2.5245098039215685e-05,
|
| 1015 |
+
"loss": 0.765,
|
| 1016 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1017 |
+
"memory_allocated (GB)": 1.46,
|
| 1018 |
"step": 50500,
|
| 1019 |
"total_memory_available (GB)": 126.62
|
| 1020 |
},
|
| 1021 |
{
|
| 1022 |
"epoch": 1.5,
|
| 1023 |
+
"grad_norm": 3.587907552719116,
|
| 1024 |
"learning_rate": 2.5e-05,
|
| 1025 |
+
"loss": 0.7608,
|
| 1026 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1027 |
+
"memory_allocated (GB)": 1.46,
|
| 1028 |
"step": 51000,
|
| 1029 |
"total_memory_available (GB)": 126.62
|
| 1030 |
},
|
| 1031 |
{
|
| 1032 |
"epoch": 1.5147058823529411,
|
| 1033 |
+
"grad_norm": 2.3503806591033936,
|
| 1034 |
"learning_rate": 2.4754901960784317e-05,
|
| 1035 |
+
"loss": 0.7619,
|
| 1036 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1037 |
+
"memory_allocated (GB)": 1.46,
|
| 1038 |
"step": 51500,
|
| 1039 |
"total_memory_available (GB)": 126.62
|
| 1040 |
},
|
| 1041 |
{
|
| 1042 |
"epoch": 1.5294117647058822,
|
| 1043 |
+
"grad_norm": 9.60110855102539,
|
| 1044 |
"learning_rate": 2.4509803921568626e-05,
|
| 1045 |
+
"loss": 0.7633,
|
| 1046 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1047 |
+
"memory_allocated (GB)": 1.46,
|
| 1048 |
"step": 52000,
|
| 1049 |
"total_memory_available (GB)": 126.62
|
| 1050 |
},
|
| 1051 |
{
|
| 1052 |
"epoch": 1.5441176470588234,
|
| 1053 |
+
"grad_norm": 9.7755708694458,
|
| 1054 |
"learning_rate": 2.4264705882352942e-05,
|
| 1055 |
+
"loss": 0.7282,
|
| 1056 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1057 |
+
"memory_allocated (GB)": 1.46,
|
| 1058 |
"step": 52500,
|
| 1059 |
"total_memory_available (GB)": 126.62
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
"epoch": 1.5588235294117647,
|
| 1063 |
+
"grad_norm": 6.872705936431885,
|
| 1064 |
"learning_rate": 2.401960784313726e-05,
|
| 1065 |
+
"loss": 0.7501,
|
| 1066 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1067 |
+
"memory_allocated (GB)": 1.46,
|
| 1068 |
"step": 53000,
|
| 1069 |
"total_memory_available (GB)": 126.62
|
| 1070 |
},
|
| 1071 |
{
|
| 1072 |
"epoch": 1.5735294117647058,
|
| 1073 |
+
"grad_norm": 4.416250228881836,
|
| 1074 |
"learning_rate": 2.3774509803921568e-05,
|
| 1075 |
+
"loss": 0.7622,
|
| 1076 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1077 |
+
"memory_allocated (GB)": 1.46,
|
| 1078 |
"step": 53500,
|
| 1079 |
"total_memory_available (GB)": 126.62
|
| 1080 |
},
|
| 1081 |
{
|
| 1082 |
"epoch": 1.5882352941176472,
|
| 1083 |
+
"grad_norm": 3.751142978668213,
|
| 1084 |
"learning_rate": 2.3529411764705884e-05,
|
| 1085 |
+
"loss": 0.7469,
|
| 1086 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1087 |
+
"memory_allocated (GB)": 1.46,
|
| 1088 |
"step": 54000,
|
| 1089 |
"total_memory_available (GB)": 126.62
|
| 1090 |
},
|
| 1091 |
{
|
| 1092 |
"epoch": 1.6029411764705883,
|
| 1093 |
+
"grad_norm": 2.0974745750427246,
|
| 1094 |
"learning_rate": 2.3284313725490196e-05,
|
| 1095 |
+
"loss": 0.7557,
|
| 1096 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1097 |
+
"memory_allocated (GB)": 1.46,
|
| 1098 |
"step": 54500,
|
| 1099 |
"total_memory_available (GB)": 126.62
|
| 1100 |
},
|
| 1101 |
{
|
| 1102 |
"epoch": 1.6176470588235294,
|
| 1103 |
+
"grad_norm": 3.7702689170837402,
|
| 1104 |
"learning_rate": 2.303921568627451e-05,
|
| 1105 |
+
"loss": 0.7468,
|
| 1106 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1107 |
+
"memory_allocated (GB)": 1.46,
|
| 1108 |
"step": 55000,
|
| 1109 |
"total_memory_available (GB)": 126.62
|
| 1110 |
},
|
| 1111 |
{
|
| 1112 |
"epoch": 1.6323529411764706,
|
| 1113 |
+
"grad_norm": 5.6461968421936035,
|
| 1114 |
"learning_rate": 2.2794117647058825e-05,
|
| 1115 |
+
"loss": 0.7501,
|
| 1116 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1117 |
+
"memory_allocated (GB)": 1.46,
|
| 1118 |
"step": 55500,
|
| 1119 |
"total_memory_available (GB)": 126.62
|
| 1120 |
},
|
| 1121 |
{
|
| 1122 |
"epoch": 1.6470588235294117,
|
| 1123 |
+
"grad_norm": 2.914978504180908,
|
| 1124 |
"learning_rate": 2.2549019607843138e-05,
|
| 1125 |
+
"loss": 0.7126,
|
| 1126 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1127 |
+
"memory_allocated (GB)": 1.46,
|
| 1128 |
"step": 56000,
|
| 1129 |
"total_memory_available (GB)": 126.62
|
| 1130 |
},
|
| 1131 |
{
|
| 1132 |
"epoch": 1.6617647058823528,
|
| 1133 |
+
"grad_norm": 5.506552219390869,
|
| 1134 |
"learning_rate": 2.2303921568627454e-05,
|
| 1135 |
+
"loss": 0.7176,
|
| 1136 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1137 |
+
"memory_allocated (GB)": 1.46,
|
| 1138 |
"step": 56500,
|
| 1139 |
"total_memory_available (GB)": 126.62
|
| 1140 |
},
|
| 1141 |
{
|
| 1142 |
"epoch": 1.6764705882352942,
|
| 1143 |
+
"grad_norm": 3.4189107418060303,
|
| 1144 |
"learning_rate": 2.2058823529411766e-05,
|
| 1145 |
+
"loss": 0.756,
|
| 1146 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1147 |
+
"memory_allocated (GB)": 1.46,
|
| 1148 |
"step": 57000,
|
| 1149 |
"total_memory_available (GB)": 126.62
|
| 1150 |
},
|
| 1151 |
{
|
| 1152 |
"epoch": 1.6911764705882353,
|
| 1153 |
+
"grad_norm": 10.335149765014648,
|
| 1154 |
"learning_rate": 2.181372549019608e-05,
|
| 1155 |
+
"loss": 0.7443,
|
| 1156 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1157 |
+
"memory_allocated (GB)": 1.46,
|
| 1158 |
"step": 57500,
|
| 1159 |
"total_memory_available (GB)": 126.62
|
| 1160 |
},
|
| 1161 |
{
|
| 1162 |
"epoch": 1.7058823529411766,
|
| 1163 |
+
"grad_norm": 1.2579532861709595,
|
| 1164 |
"learning_rate": 2.1568627450980395e-05,
|
| 1165 |
+
"loss": 0.7317,
|
| 1166 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1167 |
+
"memory_allocated (GB)": 1.46,
|
| 1168 |
"step": 58000,
|
| 1169 |
"total_memory_available (GB)": 126.62
|
| 1170 |
},
|
| 1171 |
{
|
| 1172 |
"epoch": 1.7205882352941178,
|
| 1173 |
+
"grad_norm": 5.039107322692871,
|
| 1174 |
"learning_rate": 2.1323529411764707e-05,
|
| 1175 |
+
"loss": 0.7626,
|
| 1176 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1177 |
+
"memory_allocated (GB)": 1.46,
|
| 1178 |
"step": 58500,
|
| 1179 |
"total_memory_available (GB)": 126.62
|
| 1180 |
},
|
| 1181 |
{
|
| 1182 |
"epoch": 1.7352941176470589,
|
| 1183 |
+
"grad_norm": 7.849141597747803,
|
| 1184 |
"learning_rate": 2.107843137254902e-05,
|
| 1185 |
+
"loss": 0.7506,
|
| 1186 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1187 |
+
"memory_allocated (GB)": 1.46,
|
| 1188 |
"step": 59000,
|
| 1189 |
"total_memory_available (GB)": 126.62
|
| 1190 |
},
|
| 1191 |
{
|
| 1192 |
"epoch": 1.75,
|
| 1193 |
+
"grad_norm": 5.323431491851807,
|
| 1194 |
"learning_rate": 2.0833333333333336e-05,
|
| 1195 |
+
"loss": 0.7366,
|
| 1196 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1197 |
+
"memory_allocated (GB)": 1.46,
|
| 1198 |
"step": 59500,
|
| 1199 |
"total_memory_available (GB)": 126.62
|
| 1200 |
},
|
| 1201 |
{
|
| 1202 |
"epoch": 1.7647058823529411,
|
| 1203 |
+
"grad_norm": 4.651150703430176,
|
| 1204 |
"learning_rate": 2.058823529411765e-05,
|
| 1205 |
+
"loss": 0.7467,
|
| 1206 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1207 |
+
"memory_allocated (GB)": 1.46,
|
| 1208 |
"step": 60000,
|
| 1209 |
"total_memory_available (GB)": 126.62
|
| 1210 |
},
|
| 1211 |
{
|
| 1212 |
"epoch": 1.7794117647058822,
|
| 1213 |
+
"grad_norm": 2.9276676177978516,
|
| 1214 |
"learning_rate": 2.034313725490196e-05,
|
| 1215 |
+
"loss": 0.7328,
|
| 1216 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1217 |
+
"memory_allocated (GB)": 1.46,
|
| 1218 |
"step": 60500,
|
| 1219 |
"total_memory_available (GB)": 126.62
|
| 1220 |
},
|
| 1221 |
{
|
| 1222 |
"epoch": 1.7941176470588234,
|
| 1223 |
+
"grad_norm": 6.4653401374816895,
|
| 1224 |
"learning_rate": 2.0098039215686277e-05,
|
| 1225 |
+
"loss": 0.7213,
|
| 1226 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1227 |
+
"memory_allocated (GB)": 1.46,
|
| 1228 |
"step": 61000,
|
| 1229 |
"total_memory_available (GB)": 126.62
|
| 1230 |
},
|
| 1231 |
{
|
| 1232 |
"epoch": 1.8088235294117647,
|
| 1233 |
+
"grad_norm": 5.86958122253418,
|
| 1234 |
"learning_rate": 1.9852941176470586e-05,
|
| 1235 |
+
"loss": 0.726,
|
| 1236 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1237 |
+
"memory_allocated (GB)": 1.46,
|
| 1238 |
"step": 61500,
|
| 1239 |
"total_memory_available (GB)": 126.62
|
| 1240 |
},
|
| 1241 |
{
|
| 1242 |
"epoch": 1.8235294117647058,
|
| 1243 |
+
"grad_norm": 1.54404878616333,
|
| 1244 |
"learning_rate": 1.9607843137254903e-05,
|
| 1245 |
+
"loss": 0.7228,
|
| 1246 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1247 |
+
"memory_allocated (GB)": 1.46,
|
| 1248 |
"step": 62000,
|
| 1249 |
"total_memory_available (GB)": 126.62
|
| 1250 |
},
|
| 1251 |
{
|
| 1252 |
"epoch": 1.8382352941176472,
|
| 1253 |
+
"grad_norm": 3.6029295921325684,
|
| 1254 |
"learning_rate": 1.936274509803922e-05,
|
| 1255 |
+
"loss": 0.7194,
|
| 1256 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1257 |
+
"memory_allocated (GB)": 1.46,
|
| 1258 |
"step": 62500,
|
| 1259 |
"total_memory_available (GB)": 126.62
|
| 1260 |
},
|
| 1261 |
{
|
| 1262 |
"epoch": 1.8529411764705883,
|
| 1263 |
+
"grad_norm": 13.423847198486328,
|
| 1264 |
"learning_rate": 1.9117647058823528e-05,
|
| 1265 |
+
"loss": 0.745,
|
| 1266 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1267 |
+
"memory_allocated (GB)": 1.46,
|
| 1268 |
"step": 63000,
|
| 1269 |
"total_memory_available (GB)": 126.62
|
| 1270 |
},
|
| 1271 |
{
|
| 1272 |
"epoch": 1.8676470588235294,
|
| 1273 |
+
"grad_norm": 2.904106855392456,
|
| 1274 |
"learning_rate": 1.8872549019607844e-05,
|
| 1275 |
+
"loss": 0.7321,
|
| 1276 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1277 |
+
"memory_allocated (GB)": 1.46,
|
| 1278 |
"step": 63500,
|
| 1279 |
"total_memory_available (GB)": 126.62
|
| 1280 |
},
|
| 1281 |
{
|
| 1282 |
"epoch": 1.8823529411764706,
|
| 1283 |
+
"grad_norm": 2.838707447052002,
|
| 1284 |
"learning_rate": 1.862745098039216e-05,
|
| 1285 |
+
"loss": 0.7012,
|
| 1286 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1287 |
+
"memory_allocated (GB)": 1.46,
|
| 1288 |
"step": 64000,
|
| 1289 |
"total_memory_available (GB)": 126.62
|
| 1290 |
},
|
| 1291 |
{
|
| 1292 |
"epoch": 1.8970588235294117,
|
| 1293 |
+
"grad_norm": 3.0846309661865234,
|
| 1294 |
"learning_rate": 1.8382352941176472e-05,
|
| 1295 |
+
"loss": 0.6912,
|
| 1296 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1297 |
+
"memory_allocated (GB)": 1.46,
|
| 1298 |
"step": 64500,
|
| 1299 |
"total_memory_available (GB)": 126.62
|
| 1300 |
},
|
| 1301 |
{
|
| 1302 |
"epoch": 1.9117647058823528,
|
| 1303 |
+
"grad_norm": 5.948828220367432,
|
| 1304 |
"learning_rate": 1.8137254901960785e-05,
|
| 1305 |
+
"loss": 0.6854,
|
| 1306 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1307 |
+
"memory_allocated (GB)": 1.46,
|
| 1308 |
"step": 65000,
|
| 1309 |
"total_memory_available (GB)": 126.62
|
| 1310 |
},
|
| 1311 |
{
|
| 1312 |
"epoch": 1.9264705882352942,
|
| 1313 |
+
"grad_norm": 13.732516288757324,
|
| 1314 |
"learning_rate": 1.7892156862745098e-05,
|
| 1315 |
+
"loss": 0.6873,
|
| 1316 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1317 |
+
"memory_allocated (GB)": 1.46,
|
| 1318 |
"step": 65500,
|
| 1319 |
"total_memory_available (GB)": 126.62
|
| 1320 |
},
|
| 1321 |
{
|
| 1322 |
"epoch": 1.9411764705882353,
|
| 1323 |
+
"grad_norm": 10.169962882995605,
|
| 1324 |
"learning_rate": 1.7647058823529414e-05,
|
| 1325 |
+
"loss": 0.7158,
|
| 1326 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1327 |
+
"memory_allocated (GB)": 1.46,
|
| 1328 |
"step": 66000,
|
| 1329 |
"total_memory_available (GB)": 126.62
|
| 1330 |
},
|
| 1331 |
{
|
| 1332 |
"epoch": 1.9558823529411766,
|
| 1333 |
+
"grad_norm": 8.26551628112793,
|
| 1334 |
"learning_rate": 1.7401960784313726e-05,
|
| 1335 |
+
"loss": 0.7263,
|
| 1336 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1337 |
+
"memory_allocated (GB)": 1.46,
|
| 1338 |
"step": 66500,
|
| 1339 |
"total_memory_available (GB)": 126.62
|
| 1340 |
},
|
| 1341 |
{
|
| 1342 |
"epoch": 1.9705882352941178,
|
| 1343 |
+
"grad_norm": 8.526754379272461,
|
| 1344 |
"learning_rate": 1.715686274509804e-05,
|
| 1345 |
+
"loss": 0.6879,
|
| 1346 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1347 |
+
"memory_allocated (GB)": 1.46,
|
| 1348 |
"step": 67000,
|
| 1349 |
"total_memory_available (GB)": 126.62
|
| 1350 |
},
|
| 1351 |
{
|
| 1352 |
"epoch": 1.9852941176470589,
|
| 1353 |
+
"grad_norm": 11.833305358886719,
|
| 1354 |
"learning_rate": 1.6911764705882355e-05,
|
| 1355 |
+
"loss": 0.6696,
|
| 1356 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1357 |
+
"memory_allocated (GB)": 1.46,
|
| 1358 |
"step": 67500,
|
| 1359 |
"total_memory_available (GB)": 126.62
|
| 1360 |
},
|
| 1361 |
{
|
| 1362 |
"epoch": 2.0,
|
| 1363 |
+
"grad_norm": 3.8232524394989014,
|
| 1364 |
"learning_rate": 1.6666666666666667e-05,
|
| 1365 |
+
"loss": 0.6767,
|
| 1366 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1367 |
+
"memory_allocated (GB)": 1.46,
|
| 1368 |
"step": 68000,
|
| 1369 |
"total_memory_available (GB)": 126.62
|
| 1370 |
},
|
| 1371 |
{
|
| 1372 |
"epoch": 2.014705882352941,
|
| 1373 |
+
"grad_norm": 7.861486434936523,
|
| 1374 |
"learning_rate": 1.642156862745098e-05,
|
| 1375 |
+
"loss": 0.688,
|
| 1376 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1377 |
+
"memory_allocated (GB)": 1.46,
|
| 1378 |
"step": 68500,
|
| 1379 |
"total_memory_available (GB)": 126.62
|
| 1380 |
},
|
| 1381 |
{
|
| 1382 |
"epoch": 2.0294117647058822,
|
| 1383 |
+
"grad_norm": 5.00877046585083,
|
| 1384 |
"learning_rate": 1.6176470588235296e-05,
|
| 1385 |
+
"loss": 0.6502,
|
| 1386 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1387 |
+
"memory_allocated (GB)": 1.46,
|
| 1388 |
"step": 69000,
|
| 1389 |
"total_memory_available (GB)": 126.62
|
| 1390 |
},
|
| 1391 |
{
|
| 1392 |
"epoch": 2.0441176470588234,
|
| 1393 |
+
"grad_norm": 3.252054214477539,
|
| 1394 |
"learning_rate": 1.593137254901961e-05,
|
| 1395 |
+
"loss": 0.6735,
|
| 1396 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1397 |
+
"memory_allocated (GB)": 1.46,
|
| 1398 |
"step": 69500,
|
| 1399 |
"total_memory_available (GB)": 126.62
|
| 1400 |
},
|
| 1401 |
{
|
| 1402 |
"epoch": 2.0588235294117645,
|
| 1403 |
+
"grad_norm": 8.655946731567383,
|
| 1404 |
"learning_rate": 1.568627450980392e-05,
|
| 1405 |
+
"loss": 0.7076,
|
| 1406 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1407 |
+
"memory_allocated (GB)": 1.46,
|
| 1408 |
"step": 70000,
|
| 1409 |
"total_memory_available (GB)": 126.62
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
"epoch": 2.073529411764706,
|
| 1413 |
+
"grad_norm": 5.065808296203613,
|
| 1414 |
"learning_rate": 1.5441176470588237e-05,
|
| 1415 |
+
"loss": 0.6808,
|
| 1416 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1417 |
+
"memory_allocated (GB)": 1.46,
|
| 1418 |
"step": 70500,
|
| 1419 |
"total_memory_available (GB)": 126.62
|
| 1420 |
},
|
| 1421 |
{
|
| 1422 |
"epoch": 2.088235294117647,
|
| 1423 |
+
"grad_norm": 7.614666938781738,
|
| 1424 |
"learning_rate": 1.5196078431372548e-05,
|
| 1425 |
+
"loss": 0.6565,
|
| 1426 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1427 |
+
"memory_allocated (GB)": 1.46,
|
| 1428 |
"step": 71000,
|
| 1429 |
"total_memory_available (GB)": 126.62
|
| 1430 |
},
|
| 1431 |
{
|
| 1432 |
"epoch": 2.1029411764705883,
|
| 1433 |
+
"grad_norm": 5.335823059082031,
|
| 1434 |
"learning_rate": 1.4950980392156863e-05,
|
| 1435 |
+
"loss": 0.6929,
|
| 1436 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1437 |
+
"memory_allocated (GB)": 1.46,
|
| 1438 |
"step": 71500,
|
| 1439 |
"total_memory_available (GB)": 126.62
|
| 1440 |
},
|
| 1441 |
{
|
| 1442 |
"epoch": 2.1176470588235294,
|
| 1443 |
+
"grad_norm": 8.399900436401367,
|
| 1444 |
"learning_rate": 1.4705882352941177e-05,
|
| 1445 |
+
"loss": 0.6575,
|
| 1446 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1447 |
+
"memory_allocated (GB)": 1.46,
|
| 1448 |
"step": 72000,
|
| 1449 |
"total_memory_available (GB)": 126.62
|
| 1450 |
},
|
| 1451 |
{
|
| 1452 |
"epoch": 2.1323529411764706,
|
| 1453 |
+
"grad_norm": 4.3644537925720215,
|
| 1454 |
"learning_rate": 1.4460784313725493e-05,
|
| 1455 |
+
"loss": 0.7093,
|
| 1456 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1457 |
+
"memory_allocated (GB)": 1.46,
|
| 1458 |
"step": 72500,
|
| 1459 |
"total_memory_available (GB)": 126.62
|
| 1460 |
},
|
| 1461 |
{
|
| 1462 |
"epoch": 2.1470588235294117,
|
| 1463 |
+
"grad_norm": 5.606255054473877,
|
| 1464 |
"learning_rate": 1.4215686274509804e-05,
|
| 1465 |
+
"loss": 0.6497,
|
| 1466 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1467 |
+
"memory_allocated (GB)": 1.46,
|
| 1468 |
"step": 73000,
|
| 1469 |
"total_memory_available (GB)": 126.62
|
| 1470 |
},
|
| 1471 |
{
|
| 1472 |
"epoch": 2.161764705882353,
|
| 1473 |
+
"grad_norm": 8.045038223266602,
|
| 1474 |
"learning_rate": 1.3970588235294118e-05,
|
| 1475 |
+
"loss": 0.7041,
|
| 1476 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1477 |
+
"memory_allocated (GB)": 1.46,
|
| 1478 |
"step": 73500,
|
| 1479 |
"total_memory_available (GB)": 126.62
|
| 1480 |
},
|
| 1481 |
{
|
| 1482 |
"epoch": 2.176470588235294,
|
| 1483 |
+
"grad_norm": 10.108804702758789,
|
| 1484 |
"learning_rate": 1.3725490196078432e-05,
|
| 1485 |
+
"loss": 0.6512,
|
| 1486 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1487 |
+
"memory_allocated (GB)": 1.46,
|
| 1488 |
"step": 74000,
|
| 1489 |
"total_memory_available (GB)": 126.62
|
| 1490 |
},
|
| 1491 |
{
|
| 1492 |
"epoch": 2.1911764705882355,
|
| 1493 |
+
"grad_norm": 4.9611992835998535,
|
| 1494 |
"learning_rate": 1.3480392156862745e-05,
|
| 1495 |
+
"loss": 0.6748,
|
| 1496 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1497 |
+
"memory_allocated (GB)": 1.46,
|
| 1498 |
"step": 74500,
|
| 1499 |
"total_memory_available (GB)": 126.62
|
| 1500 |
},
|
| 1501 |
{
|
| 1502 |
"epoch": 2.2058823529411766,
|
| 1503 |
+
"grad_norm": 2.3439674377441406,
|
| 1504 |
"learning_rate": 1.323529411764706e-05,
|
| 1505 |
+
"loss": 0.6671,
|
| 1506 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1507 |
+
"memory_allocated (GB)": 1.46,
|
| 1508 |
"step": 75000,
|
| 1509 |
"total_memory_available (GB)": 126.62
|
| 1510 |
},
|
| 1511 |
{
|
| 1512 |
"epoch": 2.2205882352941178,
|
| 1513 |
+
"grad_norm": 5.686610698699951,
|
| 1514 |
"learning_rate": 1.2990196078431374e-05,
|
| 1515 |
+
"loss": 0.6246,
|
| 1516 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1517 |
+
"memory_allocated (GB)": 1.46,
|
| 1518 |
"step": 75500,
|
| 1519 |
"total_memory_available (GB)": 126.62
|
| 1520 |
},
|
| 1521 |
{
|
| 1522 |
"epoch": 2.235294117647059,
|
| 1523 |
+
"grad_norm": 3.122039794921875,
|
| 1524 |
"learning_rate": 1.2745098039215686e-05,
|
| 1525 |
+
"loss": 0.7125,
|
| 1526 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1527 |
+
"memory_allocated (GB)": 1.46,
|
| 1528 |
"step": 76000,
|
| 1529 |
"total_memory_available (GB)": 126.62
|
| 1530 |
},
|
| 1531 |
{
|
| 1532 |
"epoch": 2.25,
|
| 1533 |
+
"grad_norm": 12.2803373336792,
|
| 1534 |
"learning_rate": 1.25e-05,
|
| 1535 |
+
"loss": 0.6599,
|
| 1536 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1537 |
+
"memory_allocated (GB)": 1.46,
|
| 1538 |
"step": 76500,
|
| 1539 |
"total_memory_available (GB)": 126.62
|
| 1540 |
},
|
| 1541 |
{
|
| 1542 |
"epoch": 2.264705882352941,
|
| 1543 |
+
"grad_norm": 6.284506797790527,
|
| 1544 |
"learning_rate": 1.2254901960784313e-05,
|
| 1545 |
+
"loss": 0.6759,
|
| 1546 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1547 |
+
"memory_allocated (GB)": 1.46,
|
| 1548 |
"step": 77000,
|
| 1549 |
"total_memory_available (GB)": 126.62
|
| 1550 |
},
|
| 1551 |
{
|
| 1552 |
"epoch": 2.2794117647058822,
|
| 1553 |
+
"grad_norm": 3.540226697921753,
|
| 1554 |
"learning_rate": 1.200980392156863e-05,
|
| 1555 |
+
"loss": 0.6832,
|
| 1556 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1557 |
+
"memory_allocated (GB)": 1.46,
|
| 1558 |
"step": 77500,
|
| 1559 |
"total_memory_available (GB)": 126.62
|
| 1560 |
},
|
| 1561 |
{
|
| 1562 |
"epoch": 2.2941176470588234,
|
| 1563 |
+
"grad_norm": 7.354804992675781,
|
| 1564 |
"learning_rate": 1.1764705882352942e-05,
|
| 1565 |
+
"loss": 0.6289,
|
| 1566 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1567 |
+
"memory_allocated (GB)": 1.46,
|
| 1568 |
"step": 78000,
|
| 1569 |
"total_memory_available (GB)": 126.62
|
| 1570 |
},
|
| 1571 |
{
|
| 1572 |
"epoch": 2.3088235294117645,
|
| 1573 |
+
"grad_norm": 4.479936122894287,
|
| 1574 |
"learning_rate": 1.1519607843137254e-05,
|
| 1575 |
+
"loss": 0.6235,
|
| 1576 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1577 |
+
"memory_allocated (GB)": 1.46,
|
| 1578 |
"step": 78500,
|
| 1579 |
"total_memory_available (GB)": 126.62
|
| 1580 |
},
|
| 1581 |
{
|
| 1582 |
"epoch": 2.323529411764706,
|
| 1583 |
+
"grad_norm": 5.0150299072265625,
|
| 1584 |
"learning_rate": 1.1274509803921569e-05,
|
| 1585 |
+
"loss": 0.6257,
|
| 1586 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1587 |
+
"memory_allocated (GB)": 1.46,
|
| 1588 |
"step": 79000,
|
| 1589 |
"total_memory_available (GB)": 126.62
|
| 1590 |
},
|
| 1591 |
{
|
| 1592 |
"epoch": 2.338235294117647,
|
| 1593 |
+
"grad_norm": 6.954928874969482,
|
| 1594 |
"learning_rate": 1.1029411764705883e-05,
|
| 1595 |
+
"loss": 0.6543,
|
| 1596 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1597 |
+
"memory_allocated (GB)": 1.46,
|
| 1598 |
"step": 79500,
|
| 1599 |
"total_memory_available (GB)": 126.62
|
| 1600 |
},
|
| 1601 |
{
|
| 1602 |
"epoch": 2.3529411764705883,
|
| 1603 |
+
"grad_norm": 1.8401598930358887,
|
| 1604 |
"learning_rate": 1.0784313725490197e-05,
|
| 1605 |
+
"loss": 0.6433,
|
| 1606 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1607 |
+
"memory_allocated (GB)": 1.46,
|
| 1608 |
"step": 80000,
|
| 1609 |
"total_memory_available (GB)": 126.62
|
| 1610 |
},
|
| 1611 |
{
|
| 1612 |
"epoch": 2.3676470588235294,
|
| 1613 |
+
"grad_norm": 0.5422260761260986,
|
| 1614 |
"learning_rate": 1.053921568627451e-05,
|
| 1615 |
+
"loss": 0.652,
|
| 1616 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1617 |
+
"memory_allocated (GB)": 1.46,
|
| 1618 |
"step": 80500,
|
| 1619 |
"total_memory_available (GB)": 126.62
|
| 1620 |
},
|
| 1621 |
{
|
| 1622 |
"epoch": 2.3823529411764706,
|
| 1623 |
+
"grad_norm": 6.412265300750732,
|
| 1624 |
"learning_rate": 1.0294117647058824e-05,
|
| 1625 |
+
"loss": 0.6637,
|
| 1626 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1627 |
+
"memory_allocated (GB)": 1.46,
|
| 1628 |
"step": 81000,
|
| 1629 |
"total_memory_available (GB)": 126.62
|
| 1630 |
},
|
| 1631 |
{
|
| 1632 |
"epoch": 2.3970588235294117,
|
| 1633 |
+
"grad_norm": 5.444230556488037,
|
| 1634 |
"learning_rate": 1.0049019607843139e-05,
|
| 1635 |
+
"loss": 0.6218,
|
| 1636 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1637 |
+
"memory_allocated (GB)": 1.46,
|
| 1638 |
"step": 81500,
|
| 1639 |
"total_memory_available (GB)": 126.62
|
| 1640 |
},
|
| 1641 |
{
|
| 1642 |
"epoch": 2.411764705882353,
|
| 1643 |
+
"grad_norm": 3.0248329639434814,
|
| 1644 |
"learning_rate": 9.803921568627451e-06,
|
| 1645 |
+
"loss": 0.6663,
|
| 1646 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1647 |
+
"memory_allocated (GB)": 1.46,
|
| 1648 |
"step": 82000,
|
| 1649 |
"total_memory_available (GB)": 126.62
|
| 1650 |
},
|
| 1651 |
{
|
| 1652 |
"epoch": 2.426470588235294,
|
| 1653 |
+
"grad_norm": 2.8557262420654297,
|
| 1654 |
"learning_rate": 9.558823529411764e-06,
|
| 1655 |
+
"loss": 0.6768,
|
| 1656 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1657 |
+
"memory_allocated (GB)": 1.46,
|
| 1658 |
"step": 82500,
|
| 1659 |
"total_memory_available (GB)": 126.62
|
| 1660 |
},
|
| 1661 |
{
|
| 1662 |
"epoch": 2.4411764705882355,
|
| 1663 |
+
"grad_norm": 3.229672908782959,
|
| 1664 |
"learning_rate": 9.31372549019608e-06,
|
| 1665 |
+
"loss": 0.6566,
|
| 1666 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1667 |
+
"memory_allocated (GB)": 1.46,
|
| 1668 |
"step": 83000,
|
| 1669 |
"total_memory_available (GB)": 126.62
|
| 1670 |
},
|
| 1671 |
{
|
| 1672 |
"epoch": 2.4558823529411766,
|
| 1673 |
+
"grad_norm": 6.890370845794678,
|
| 1674 |
"learning_rate": 9.068627450980392e-06,
|
| 1675 |
+
"loss": 0.6589,
|
| 1676 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1677 |
+
"memory_allocated (GB)": 1.46,
|
| 1678 |
"step": 83500,
|
| 1679 |
"total_memory_available (GB)": 126.62
|
| 1680 |
},
|
| 1681 |
{
|
| 1682 |
"epoch": 2.4705882352941178,
|
| 1683 |
+
"grad_norm": 7.899547100067139,
|
| 1684 |
"learning_rate": 8.823529411764707e-06,
|
| 1685 |
+
"loss": 0.6775,
|
| 1686 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1687 |
+
"memory_allocated (GB)": 1.46,
|
| 1688 |
"step": 84000,
|
| 1689 |
"total_memory_available (GB)": 126.62
|
| 1690 |
},
|
| 1691 |
{
|
| 1692 |
"epoch": 2.485294117647059,
|
| 1693 |
+
"grad_norm": 7.249077796936035,
|
| 1694 |
"learning_rate": 8.57843137254902e-06,
|
| 1695 |
+
"loss": 0.6572,
|
| 1696 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1697 |
+
"memory_allocated (GB)": 1.46,
|
| 1698 |
"step": 84500,
|
| 1699 |
"total_memory_available (GB)": 126.62
|
| 1700 |
},
|
| 1701 |
{
|
| 1702 |
"epoch": 2.5,
|
| 1703 |
+
"grad_norm": 5.247090816497803,
|
| 1704 |
"learning_rate": 8.333333333333334e-06,
|
| 1705 |
+
"loss": 0.6348,
|
| 1706 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1707 |
+
"memory_allocated (GB)": 1.46,
|
| 1708 |
"step": 85000,
|
| 1709 |
"total_memory_available (GB)": 126.62
|
| 1710 |
},
|
| 1711 |
{
|
| 1712 |
"epoch": 2.514705882352941,
|
| 1713 |
+
"grad_norm": 5.179773330688477,
|
| 1714 |
"learning_rate": 8.088235294117648e-06,
|
| 1715 |
+
"loss": 0.6224,
|
| 1716 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1717 |
+
"memory_allocated (GB)": 1.46,
|
| 1718 |
"step": 85500,
|
| 1719 |
"total_memory_available (GB)": 126.62
|
| 1720 |
},
|
| 1721 |
{
|
| 1722 |
"epoch": 2.5294117647058822,
|
| 1723 |
+
"grad_norm": 8.414857864379883,
|
| 1724 |
"learning_rate": 7.84313725490196e-06,
|
| 1725 |
+
"loss": 0.6136,
|
| 1726 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1727 |
+
"memory_allocated (GB)": 1.46,
|
| 1728 |
"step": 86000,
|
| 1729 |
"total_memory_available (GB)": 126.62
|
| 1730 |
},
|
| 1731 |
{
|
| 1732 |
"epoch": 2.5441176470588234,
|
| 1733 |
+
"grad_norm": 1.9255198240280151,
|
| 1734 |
"learning_rate": 7.598039215686274e-06,
|
| 1735 |
+
"loss": 0.6418,
|
| 1736 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1737 |
+
"memory_allocated (GB)": 1.46,
|
| 1738 |
"step": 86500,
|
| 1739 |
"total_memory_available (GB)": 126.62
|
| 1740 |
},
|
| 1741 |
{
|
| 1742 |
"epoch": 2.5588235294117645,
|
| 1743 |
+
"grad_norm": 4.827670574188232,
|
| 1744 |
"learning_rate": 7.3529411764705884e-06,
|
| 1745 |
+
"loss": 0.6562,
|
| 1746 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1747 |
+
"memory_allocated (GB)": 1.46,
|
| 1748 |
"step": 87000,
|
| 1749 |
"total_memory_available (GB)": 126.62
|
| 1750 |
},
|
| 1751 |
{
|
| 1752 |
"epoch": 2.5735294117647056,
|
| 1753 |
+
"grad_norm": 5.845033645629883,
|
| 1754 |
"learning_rate": 7.107843137254902e-06,
|
| 1755 |
+
"loss": 0.6471,
|
| 1756 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1757 |
+
"memory_allocated (GB)": 1.46,
|
| 1758 |
"step": 87500,
|
| 1759 |
"total_memory_available (GB)": 126.62
|
| 1760 |
},
|
| 1761 |
{
|
| 1762 |
"epoch": 2.588235294117647,
|
| 1763 |
+
"grad_norm": 6.777614593505859,
|
| 1764 |
"learning_rate": 6.862745098039216e-06,
|
| 1765 |
+
"loss": 0.6311,
|
| 1766 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1767 |
+
"memory_allocated (GB)": 1.46,
|
| 1768 |
"step": 88000,
|
| 1769 |
"total_memory_available (GB)": 126.62
|
| 1770 |
},
|
| 1771 |
{
|
| 1772 |
"epoch": 2.6029411764705883,
|
| 1773 |
+
"grad_norm": 11.79529094696045,
|
| 1774 |
"learning_rate": 6.61764705882353e-06,
|
| 1775 |
+
"loss": 0.6289,
|
| 1776 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1777 |
+
"memory_allocated (GB)": 1.46,
|
| 1778 |
"step": 88500,
|
| 1779 |
"total_memory_available (GB)": 126.62
|
| 1780 |
},
|
| 1781 |
{
|
| 1782 |
"epoch": 2.6176470588235294,
|
| 1783 |
+
"grad_norm": 8.824967384338379,
|
| 1784 |
"learning_rate": 6.372549019607843e-06,
|
| 1785 |
+
"loss": 0.619,
|
| 1786 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1787 |
+
"memory_allocated (GB)": 1.46,
|
| 1788 |
"step": 89000,
|
| 1789 |
"total_memory_available (GB)": 126.62
|
| 1790 |
},
|
| 1791 |
{
|
| 1792 |
"epoch": 2.6323529411764706,
|
| 1793 |
+
"grad_norm": 6.913949012756348,
|
| 1794 |
"learning_rate": 6.127450980392157e-06,
|
| 1795 |
+
"loss": 0.652,
|
| 1796 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1797 |
+
"memory_allocated (GB)": 1.46,
|
| 1798 |
"step": 89500,
|
| 1799 |
"total_memory_available (GB)": 126.62
|
| 1800 |
},
|
| 1801 |
{
|
| 1802 |
"epoch": 2.6470588235294117,
|
| 1803 |
+
"grad_norm": 3.8313751220703125,
|
| 1804 |
"learning_rate": 5.882352941176471e-06,
|
| 1805 |
+
"loss": 0.6701,
|
| 1806 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1807 |
+
"memory_allocated (GB)": 1.46,
|
| 1808 |
"step": 90000,
|
| 1809 |
"total_memory_available (GB)": 126.62
|
| 1810 |
},
|
| 1811 |
{
|
| 1812 |
"epoch": 2.661764705882353,
|
| 1813 |
+
"grad_norm": 8.215780258178711,
|
| 1814 |
"learning_rate": 5.637254901960784e-06,
|
| 1815 |
+
"loss": 0.6433,
|
| 1816 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1817 |
+
"memory_allocated (GB)": 1.46,
|
| 1818 |
"step": 90500,
|
| 1819 |
"total_memory_available (GB)": 126.62
|
| 1820 |
},
|
| 1821 |
{
|
| 1822 |
"epoch": 2.6764705882352944,
|
| 1823 |
+
"grad_norm": 6.524360179901123,
|
| 1824 |
"learning_rate": 5.392156862745099e-06,
|
| 1825 |
+
"loss": 0.6048,
|
| 1826 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1827 |
+
"memory_allocated (GB)": 1.46,
|
| 1828 |
"step": 91000,
|
| 1829 |
"total_memory_available (GB)": 126.62
|
| 1830 |
},
|
| 1831 |
{
|
| 1832 |
"epoch": 2.6911764705882355,
|
| 1833 |
+
"grad_norm": 15.196798324584961,
|
| 1834 |
"learning_rate": 5.147058823529412e-06,
|
| 1835 |
+
"loss": 0.6427,
|
| 1836 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1837 |
+
"memory_allocated (GB)": 1.46,
|
| 1838 |
"step": 91500,
|
| 1839 |
"total_memory_available (GB)": 126.62
|
| 1840 |
},
|
| 1841 |
{
|
| 1842 |
"epoch": 2.7058823529411766,
|
| 1843 |
+
"grad_norm": 1.6884765625,
|
| 1844 |
"learning_rate": 4.901960784313726e-06,
|
| 1845 |
+
"loss": 0.6236,
|
| 1846 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1847 |
+
"memory_allocated (GB)": 1.46,
|
| 1848 |
"step": 92000,
|
| 1849 |
"total_memory_available (GB)": 126.62
|
| 1850 |
},
|
| 1851 |
{
|
| 1852 |
"epoch": 2.7205882352941178,
|
| 1853 |
+
"grad_norm": 2.3077046871185303,
|
| 1854 |
"learning_rate": 4.65686274509804e-06,
|
| 1855 |
+
"loss": 0.6163,
|
| 1856 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1857 |
+
"memory_allocated (GB)": 1.46,
|
| 1858 |
"step": 92500,
|
| 1859 |
"total_memory_available (GB)": 126.62
|
| 1860 |
},
|
| 1861 |
{
|
| 1862 |
"epoch": 2.735294117647059,
|
| 1863 |
+
"grad_norm": 5.261138916015625,
|
| 1864 |
"learning_rate": 4.411764705882353e-06,
|
| 1865 |
+
"loss": 0.6361,
|
| 1866 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1867 |
+
"memory_allocated (GB)": 1.46,
|
| 1868 |
"step": 93000,
|
| 1869 |
"total_memory_available (GB)": 126.62
|
| 1870 |
},
|
| 1871 |
{
|
| 1872 |
"epoch": 2.75,
|
| 1873 |
+
"grad_norm": 6.308595657348633,
|
| 1874 |
"learning_rate": 4.166666666666667e-06,
|
| 1875 |
+
"loss": 0.6245,
|
| 1876 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1877 |
+
"memory_allocated (GB)": 1.46,
|
| 1878 |
"step": 93500,
|
| 1879 |
"total_memory_available (GB)": 126.62
|
| 1880 |
},
|
| 1881 |
{
|
| 1882 |
"epoch": 2.764705882352941,
|
| 1883 |
+
"grad_norm": 5.467650413513184,
|
| 1884 |
"learning_rate": 3.92156862745098e-06,
|
| 1885 |
+
"loss": 0.621,
|
| 1886 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1887 |
+
"memory_allocated (GB)": 1.46,
|
| 1888 |
"step": 94000,
|
| 1889 |
"total_memory_available (GB)": 126.62
|
| 1890 |
},
|
| 1891 |
{
|
| 1892 |
"epoch": 2.7794117647058822,
|
| 1893 |
+
"grad_norm": 5.5255632400512695,
|
| 1894 |
"learning_rate": 3.6764705882352942e-06,
|
| 1895 |
+
"loss": 0.6446,
|
| 1896 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1897 |
+
"memory_allocated (GB)": 1.46,
|
| 1898 |
"step": 94500,
|
| 1899 |
"total_memory_available (GB)": 126.62
|
| 1900 |
},
|
| 1901 |
{
|
| 1902 |
"epoch": 2.7941176470588234,
|
| 1903 |
+
"grad_norm": 9.804576873779297,
|
| 1904 |
"learning_rate": 3.431372549019608e-06,
|
| 1905 |
+
"loss": 0.6226,
|
| 1906 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1907 |
+
"memory_allocated (GB)": 1.46,
|
| 1908 |
"step": 95000,
|
| 1909 |
"total_memory_available (GB)": 126.62
|
| 1910 |
},
|
| 1911 |
{
|
| 1912 |
"epoch": 2.8088235294117645,
|
| 1913 |
+
"grad_norm": 7.767389297485352,
|
| 1914 |
"learning_rate": 3.1862745098039216e-06,
|
| 1915 |
+
"loss": 0.6147,
|
| 1916 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1917 |
+
"memory_allocated (GB)": 1.46,
|
| 1918 |
"step": 95500,
|
| 1919 |
"total_memory_available (GB)": 126.62
|
| 1920 |
},
|
| 1921 |
{
|
| 1922 |
"epoch": 2.8235294117647056,
|
| 1923 |
+
"grad_norm": 2.8358209133148193,
|
| 1924 |
"learning_rate": 2.9411764705882355e-06,
|
| 1925 |
+
"loss": 0.6416,
|
| 1926 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1927 |
+
"memory_allocated (GB)": 1.46,
|
| 1928 |
"step": 96000,
|
| 1929 |
"total_memory_available (GB)": 126.62
|
| 1930 |
},
|
| 1931 |
{
|
| 1932 |
"epoch": 2.838235294117647,
|
| 1933 |
+
"grad_norm": 7.5477190017700195,
|
| 1934 |
"learning_rate": 2.6960784313725493e-06,
|
| 1935 |
+
"loss": 0.6328,
|
| 1936 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1937 |
+
"memory_allocated (GB)": 1.46,
|
| 1938 |
"step": 96500,
|
| 1939 |
"total_memory_available (GB)": 126.62
|
| 1940 |
},
|
| 1941 |
{
|
| 1942 |
"epoch": 2.8529411764705883,
|
| 1943 |
+
"grad_norm": 13.355735778808594,
|
| 1944 |
"learning_rate": 2.450980392156863e-06,
|
| 1945 |
+
"loss": 0.6087,
|
| 1946 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1947 |
+
"memory_allocated (GB)": 1.46,
|
| 1948 |
"step": 97000,
|
| 1949 |
"total_memory_available (GB)": 126.62
|
| 1950 |
},
|
| 1951 |
{
|
| 1952 |
"epoch": 2.8676470588235294,
|
| 1953 |
+
"grad_norm": 5.399519443511963,
|
| 1954 |
"learning_rate": 2.2058823529411767e-06,
|
| 1955 |
+
"loss": 0.6063,
|
| 1956 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1957 |
+
"memory_allocated (GB)": 1.46,
|
| 1958 |
"step": 97500,
|
| 1959 |
"total_memory_available (GB)": 126.62
|
| 1960 |
},
|
| 1961 |
{
|
| 1962 |
"epoch": 2.8823529411764706,
|
| 1963 |
+
"grad_norm": 7.818796634674072,
|
| 1964 |
"learning_rate": 1.96078431372549e-06,
|
| 1965 |
+
"loss": 0.6211,
|
| 1966 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1967 |
+
"memory_allocated (GB)": 1.46,
|
| 1968 |
"step": 98000,
|
| 1969 |
"total_memory_available (GB)": 126.62
|
| 1970 |
},
|
| 1971 |
{
|
| 1972 |
"epoch": 2.8970588235294117,
|
| 1973 |
+
"grad_norm": 4.248461723327637,
|
| 1974 |
"learning_rate": 1.715686274509804e-06,
|
| 1975 |
+
"loss": 0.6054,
|
| 1976 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1977 |
+
"memory_allocated (GB)": 1.46,
|
| 1978 |
"step": 98500,
|
| 1979 |
"total_memory_available (GB)": 126.62
|
| 1980 |
},
|
| 1981 |
{
|
| 1982 |
"epoch": 2.911764705882353,
|
| 1983 |
+
"grad_norm": 6.80518102645874,
|
| 1984 |
"learning_rate": 1.4705882352941177e-06,
|
| 1985 |
+
"loss": 0.6261,
|
| 1986 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1987 |
+
"memory_allocated (GB)": 1.46,
|
| 1988 |
"step": 99000,
|
| 1989 |
"total_memory_available (GB)": 126.62
|
| 1990 |
},
|
| 1991 |
{
|
| 1992 |
"epoch": 2.9264705882352944,
|
| 1993 |
+
"grad_norm": 0.6795050501823425,
|
| 1994 |
"learning_rate": 1.2254901960784314e-06,
|
| 1995 |
+
"loss": 0.6062,
|
| 1996 |
+
"max_memory_allocated (GB)": 2.1,
|
| 1997 |
+
"memory_allocated (GB)": 1.46,
|
| 1998 |
"step": 99500,
|
| 1999 |
"total_memory_available (GB)": 126.62
|
| 2000 |
},
|
| 2001 |
{
|
| 2002 |
"epoch": 2.9411764705882355,
|
| 2003 |
+
"grad_norm": 6.518540382385254,
|
| 2004 |
"learning_rate": 9.80392156862745e-07,
|
| 2005 |
+
"loss": 0.6304,
|
| 2006 |
+
"max_memory_allocated (GB)": 2.1,
|
| 2007 |
+
"memory_allocated (GB)": 1.46,
|
| 2008 |
"step": 100000,
|
| 2009 |
"total_memory_available (GB)": 126.62
|
| 2010 |
},
|
| 2011 |
{
|
| 2012 |
"epoch": 2.9558823529411766,
|
| 2013 |
+
"grad_norm": 5.016809463500977,
|
| 2014 |
"learning_rate": 7.352941176470589e-07,
|
| 2015 |
+
"loss": 0.636,
|
| 2016 |
+
"max_memory_allocated (GB)": 2.1,
|
| 2017 |
+
"memory_allocated (GB)": 1.46,
|
| 2018 |
"step": 100500,
|
| 2019 |
"total_memory_available (GB)": 126.62
|
| 2020 |
},
|
| 2021 |
{
|
| 2022 |
"epoch": 2.9705882352941178,
|
| 2023 |
+
"grad_norm": 7.016627788543701,
|
| 2024 |
"learning_rate": 4.901960784313725e-07,
|
| 2025 |
+
"loss": 0.6178,
|
| 2026 |
+
"max_memory_allocated (GB)": 2.1,
|
| 2027 |
+
"memory_allocated (GB)": 1.46,
|
| 2028 |
"step": 101000,
|
| 2029 |
"total_memory_available (GB)": 126.62
|
| 2030 |
},
|
| 2031 |
{
|
| 2032 |
"epoch": 2.985294117647059,
|
| 2033 |
+
"grad_norm": 12.064628601074219,
|
| 2034 |
"learning_rate": 2.4509803921568627e-07,
|
| 2035 |
+
"loss": 0.6343,
|
| 2036 |
+
"max_memory_allocated (GB)": 2.1,
|
| 2037 |
+
"memory_allocated (GB)": 1.46,
|
| 2038 |
"step": 101500,
|
| 2039 |
"total_memory_available (GB)": 126.62
|
| 2040 |
},
|
| 2041 |
{
|
| 2042 |
"epoch": 3.0,
|
| 2043 |
+
"grad_norm": 3.9333810806274414,
|
| 2044 |
"learning_rate": 0.0,
|
| 2045 |
+
"loss": 0.5962,
|
| 2046 |
+
"max_memory_allocated (GB)": 2.1,
|
| 2047 |
+
"memory_allocated (GB)": 1.46,
|
| 2048 |
"step": 102000,
|
| 2049 |
"total_memory_available (GB)": 126.62
|
| 2050 |
},
|
| 2051 |
{
|
| 2052 |
"epoch": 3.0,
|
| 2053 |
+
"max_memory_allocated (GB)": 2.1,
|
| 2054 |
+
"memory_allocated (GB)": 1.46,
|
| 2055 |
"step": 102000,
|
| 2056 |
"total_flos": 6.324139790696448e+19,
|
| 2057 |
"total_memory_available (GB)": 126.62,
|
| 2058 |
+
"train_loss": 0.8088104343788297,
|
| 2059 |
+
"train_runtime": 4169.6612,
|
| 2060 |
+
"train_samples_per_second": 195.699,
|
| 2061 |
+
"train_steps_per_second": 24.462
|
| 2062 |
}
|
| 2063 |
],
|
| 2064 |
"logging_steps": 500,
|