Update metrics - Run 20251012_065325

best_model.json

@@ -1,36 +1,39 @@
 {
-  "model_name": "David-partial_shared-hierarchical_tree",
+  "model_name": "David-partial_shared-deep_efficiency",
   "run_id": "20251012_065325",
-  "timestamp": "2025-10-12T07:35:40.795003",
-  "best_val_acc": 75.414,
-  "best_epoch": 9,
-  "final_train_acc": 87.91102174814056,
-  "final_train_loss": 1.3676543517615467,
+  "timestamp": "2025-10-12T07:38:25.221146",
+  "best_val_acc": 81.158,
+  "best_epoch": 0,
+  "final_train_acc": 78.10129358623817,
+  "final_train_loss": 1.346226876060041,
   "scale_accuracies": {
-    "256": 74.794,
-    "512": 75.394,
-    "768": 75.398,
-    "1024": 73.42
+    "256": 81.158
   },
   "architecture": {
-    "preset": "balanced",
+    "preset": "clip_vit_l14_ultra_deep",
     "sharing_mode": "partial_shared",
-    "fusion_mode": "hierarchical_tree",
+    "fusion_mode": "deep_efficiency",
     "scales": [
       256,
       512,
       768,
-      1024
+      1024,
+      1280,
+      1536,
+      1792,
+      2048,
+      2304,
+      2560
     ],
-    "feature_dim": 512,
+    "feature_dim": 768,
     "num_classes": 1000,
     "use_belly": true,
-    "belly_expand": 2.0
+    "belly_expand": 3.0
   },
   "training": {
     "dataset": "AbstractPhil/imagenet-clip-features-orderly",
-    "model_variant": "clip_vit_laion_b32",
-    "num_epochs": 20,
+    "model_variant": "clip_vit_l14",
+    "num_epochs": 10,
     "batch_size": 1024,
     "learning_rate": 0.001,
     "rose_weight": "0.1\u21920.5",
@@ -39,10 +42,10 @@
     "scheduler": "cosine_restarts"
   },
   "files": {
-    "weights_safetensors": "weights/David-partial_shared-hierarchical_tree/20251012_065325/best_model.safetensors",
-    "weights_pytorch": "weights/David-partial_shared-hierarchical_tree/20251012_065325/best_model.pth",
-    "config": "weights/David-partial_shared-hierarchical_tree/20251012_065325/david_config.json",
-    "training_config": "weights/David-partial_shared-hierarchical_tree/20251012_065325/train_config.json",
-    "tensorboard": "runs/David-partial_shared-hierarchical_tree/20251012_065325/"
+    "weights_safetensors": "weights/David-partial_shared-deep_efficiency/20251012_065325/best_model.safetensors",
+    "weights_pytorch": "weights/David-partial_shared-deep_efficiency/20251012_065325/best_model.pth",
+    "config": "weights/David-partial_shared-deep_efficiency/20251012_065325/david_config.json",
+    "training_config": "weights/David-partial_shared-deep_efficiency/20251012_065325/train_config.json",
+    "tensorboard": "runs/David-partial_shared-deep_efficiency/20251012_065325/"
   }
 }