Update README - Run 20251104_144102

README.md

@@ -12,7 +12,7 @@ datasets:
 metrics:
 - accuracy
 model-index:
-- name: David-partial_shared-cantor_scale
+- name: David-decoupled-cantor_scale
   results:
   - task:
       type: image-classification
@@ -21,7 +21,7 @@ model-index:
       type: imagenet-1k
     metrics:
     - type: accuracy
-      value: 84.92
+      value: 84.30
 ---
 
 # David: Multi-Scale Feature Classifier
@@ -32,12 +32,12 @@ as class prototypes with role-weighted similarity computation (Rose Loss).
 ## Model Details
 
 ### Architecture
-- **Preset**: clip_vit_bigg14_cantor
-- **Sharing Mode**: partial_shared
+- **Preset**: clip_vit_bigg14_cantor_decoupled
+- **Sharing Mode**: decoupled
 - **Fusion Mode**: cantor_scale
 - **Scales**: [384, 512, 768, 1024, 1280, 1536, 1792, 2048]
 - **Feature Dim**: 1280
-- **Parameters**: 82,601,993
+- **Parameters**: 70,207,625
 
 ### Training Configuration
 - **Dataset**: AbstractPhil/imagenet-clip-features-orderly
@@ -51,19 +51,19 @@ as class prototypes with role-weighted similarity computation (Rose Loss).
 ## Performance
 
 ### Best Results
-- **Validation Accuracy**: 84.92%
-- **Best Epoch**: 4
-- **Final Train Accuracy**: 92.33%
+- **Validation Accuracy**: 84.30%
+- **Best Epoch**: 0
+- **Final Train Accuracy**: 83.51%
 
 ### Per-Scale Performance
-- **Scale 384**: 84.71%
-- **Scale 512**: 84.68%
-- **Scale 768**: 84.66%
-- **Scale 1024**: 84.70%
-- **Scale 1280**: 84.47%
-- **Scale 1536**: 84.35%
-- **Scale 1792**: 84.07%
-- **Scale 2048**: 84.28%
+- **Scale 384**: 82.25%
+- **Scale 512**: 82.68%
+- **Scale 768**: 83.39%
+- **Scale 1024**: 83.31%
+- **Scale 1280**: 83.26%
+- **Scale 1536**: 83.20%
+- **Scale 1792**: 83.19%
+- **Scale 2048**: 83.25%
 
 
 ## Usage
@@ -80,19 +80,19 @@ AbstractPhil/gated-david/
 ├── README.md                         # This file
 ├── best_model.json                   # Latest best model info
 ├── weights/
-│   └── david_clip_vit_bigg14_cantor/
-│       └── 20251104_140742/
+│   └── david_clip_vit_bigg14_cantor_decoupled/
+│       └── 20251104_144102/
 │           ├── MODEL_SUMMARY.txt     # 🎯 Human-readable performance summary
 │           ├── training_history.json # 📈 Epoch-by-epoch training curve
-│           ├── best_model_acc84.92.safetensors  # ⭐ Accuracy in filename!
-│           ├── best_model_acc84.92_metadata.json
+│           ├── best_model_acc84.30.safetensors  # ⭐ Accuracy in filename!
+│           ├── best_model_acc84.30_metadata.json
 │           ├── final_model.safetensors
 │           ├── checkpoint_epoch_X_accYY.YY.safetensors
 │           ├── david_config.json
 │           └── train_config.json
 └── runs/
-    └── david_clip_vit_bigg14_cantor/
-        └── 20251104_140742/
+    └── david_clip_vit_bigg14_cantor_decoupled/
+        └── 20251104_144102/
             └── events.out.tfevents.* # TensorBoard logs
 ```
 
@@ -105,9 +105,9 @@ from huggingface_hub import hf_hub_download
 # Browse available models in MODELS_INDEX.json first!
 
 # Specify model variant and run
-model_name = "david_clip_vit_bigg14_cantor"
-run_id = "20251104_140742"
-accuracy = "84.92"  # From MODELS_INDEX.json
+model_name = "david_clip_vit_bigg14_cantor_decoupled"
+run_id = "20251104_144102"
+accuracy = "84.30"  # From MODELS_INDEX.json
 
 # Download config
 config_path = hf_hub_download(
@@ -198,7 +198,7 @@ score = w_anchor * sim(z, anchor) + w_need * sim(z, need) + ...
   author = {AbstractPhil},
   year = {2025},
   url = {https://huggingface.co/AbstractPhil/gated-david},
-  note = {Run ID: 20251104_140742}
+  note = {Run ID: 20251104_144102}
 }
 ```
 
@@ -213,4 +213,4 @@ Special thanks to Claude (Anthropic) for debugging assistance.
 
 ---
 
-*Generated on 2025-11-04 14:34:13*
+*Generated on 2025-11-04 14:45:04*