Upload best.meta.json
Browse files
geom_multitask_mnist_fashion_cifar100_best.meta.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"global_step":
|
| 4 |
-
"avg_loss":
|
| 5 |
-
"best_metric": 0.
|
| 6 |
-
"timestamp": "2025-08-25T23:
|
| 7 |
"config": {
|
| 8 |
"seed": 42,
|
| 9 |
"device": "cuda",
|
|
@@ -170,32 +170,32 @@
|
|
| 170 |
},
|
| 171 |
"classification": {
|
| 172 |
"use_cls": true,
|
| 173 |
-
"lambda_cls":
|
| 174 |
"tau_cls": 0.35
|
| 175 |
},
|
| 176 |
"losses": {
|
| 177 |
"use_rose_mag_hinge": true,
|
| 178 |
-
"lambda_rose_mag": 0.
|
| 179 |
-
"rose_margin":
|
| 180 |
"use_dual_contrastive": true,
|
| 181 |
-
"lambda_dual":
|
| 182 |
-
"temp_dual": 0.
|
| 183 |
"use_rose_contrastive": true,
|
| 184 |
-
"lambda_rose_contrastive":
|
| 185 |
"temp_rose_contrastive": 0.5
|
| 186 |
},
|
| 187 |
"train": {
|
| 188 |
"epochs": 10,
|
| 189 |
"eval_every_epochs": 1,
|
| 190 |
-
"lr": 0.
|
| 191 |
"grad_clip_norm": 1.0,
|
| 192 |
"use_anneal": true,
|
| 193 |
"anneal": {
|
| 194 |
-
"tau_start": 0.
|
| 195 |
-
"drift_start": 0.
|
| 196 |
-
"lambda_cls_start": 0
|
| 197 |
-
"lambda_dual_start": 0.
|
| 198 |
-
"lambda_rose_contrastive_start": 0.
|
| 199 |
}
|
| 200 |
},
|
| 201 |
"io": {
|
|
@@ -228,16 +228,16 @@
|
|
| 228 |
},
|
| 229 |
"per_task": {
|
| 230 |
"mnist": {
|
| 231 |
-
"top1": 0.
|
| 232 |
-
"top3": 0.
|
| 233 |
},
|
| 234 |
"fashion": {
|
| 235 |
-
"top1": 0.
|
| 236 |
-
"top3": 0.
|
| 237 |
},
|
| 238 |
"cifar100": {
|
| 239 |
-
"top1": 0.
|
| 240 |
-
"top3": 0.
|
| 241 |
}
|
| 242 |
}
|
| 243 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 42,
|
| 3 |
+
"global_step": 29610,
|
| 4 |
+
"avg_loss": 29.386593157179814,
|
| 5 |
+
"best_metric": 0.4609666666666667,
|
| 6 |
+
"timestamp": "2025-08-25T23:11:43.698260",
|
| 7 |
"config": {
|
| 8 |
"seed": 42,
|
| 9 |
"device": "cuda",
|
|
|
|
| 170 |
},
|
| 171 |
"classification": {
|
| 172 |
"use_cls": true,
|
| 173 |
+
"lambda_cls": 9.61,
|
| 174 |
"tau_cls": 0.35
|
| 175 |
},
|
| 176 |
"losses": {
|
| 177 |
"use_rose_mag_hinge": true,
|
| 178 |
+
"lambda_rose_mag": 0.025,
|
| 179 |
+
"rose_margin": 0.2915,
|
| 180 |
"use_dual_contrastive": true,
|
| 181 |
+
"lambda_dual": 0.15,
|
| 182 |
+
"temp_dual": 0.25,
|
| 183 |
"use_rose_contrastive": true,
|
| 184 |
+
"lambda_rose_contrastive": 2.7329999999999997,
|
| 185 |
"temp_rose_contrastive": 0.5
|
| 186 |
},
|
| 187 |
"train": {
|
| 188 |
"epochs": 10,
|
| 189 |
"eval_every_epochs": 1,
|
| 190 |
+
"lr": 0.0005,
|
| 191 |
"grad_clip_norm": 1.0,
|
| 192 |
"use_anneal": true,
|
| 193 |
"anneal": {
|
| 194 |
+
"tau_start": 0.55,
|
| 195 |
+
"drift_start": 0.05,
|
| 196 |
+
"lambda_cls_start": 1.0,
|
| 197 |
+
"lambda_dual_start": 0.15,
|
| 198 |
+
"lambda_rose_contrastive_start": 0.15
|
| 199 |
}
|
| 200 |
},
|
| 201 |
"io": {
|
|
|
|
| 228 |
},
|
| 229 |
"per_task": {
|
| 230 |
"mnist": {
|
| 231 |
+
"top1": 0.7906,
|
| 232 |
+
"top3": 0.9587
|
| 233 |
},
|
| 234 |
"fashion": {
|
| 235 |
+
"top1": 0.5807,
|
| 236 |
+
"top3": 0.9087
|
| 237 |
},
|
| 238 |
"cifar100": {
|
| 239 |
+
"top1": 0.0116,
|
| 240 |
+
"top3": 0.0404
|
| 241 |
}
|
| 242 |
}
|
| 243 |
}
|