Update README - Run 20251012_152245

README.md

@@ -12,7 +12,7 @@ datasets:
 metrics:
 - accuracy
 model-index:
-- name: David-hierarchical-progressive
+- name: David-partial_shared-hierarchical_tree
   results:
   - task:
       type: image-classification
@@ -21,7 +21,7 @@ model-index:
       type: imagenet-1k
     metrics:
     - type: accuracy
-      value: 72.23
+      value: 73.29
 ---
 
 # David: Multi-Scale Crystal Classifier
@@ -32,12 +32,12 @@ as class prototypes with role-weighted similarity computation (Rose Loss).
 ## Model Details
 
 ### Architecture
-- **Preset**: hierarchical_refinement
-- **Sharing Mode**: hierarchical
-- **Fusion Mode**: progressive
+- **Preset**: balanced
+- **Sharing Mode**: partial_shared
+- **Fusion Mode**: hierarchical_tree
 - **Scales**: [256, 512, 768, 1024]
 - **Feature Dim**: 512
-- **Parameters**: 10,493,184
+- **Parameters**: 8,758,271
 
 ### Training Configuration
 - **Dataset**: AbstractPhil/imagenet-clip-features-orderly
@@ -51,15 +51,15 @@ as class prototypes with role-weighted similarity computation (Rose Loss).
 ## Performance
 
 ### Best Results
-- **Validation Accuracy**: 72.23%
+- **Validation Accuracy**: 73.29%
 - **Best Epoch**: 0
-- **Final Train Accuracy**: 68.42%
+- **Final Train Accuracy**: 70.12%
 
 ### Per-Scale Performance
-- **Scale 256**: 65.87%
-- **Scale 512**: 69.96%
-- **Scale 768**: 71.41%
-- **Scale 1024**: 71.52%
+- **Scale 256**: 71.33%
+- **Scale 512**: 72.45%
+- **Scale 768**: 72.26%
+- **Scale 1024**: 73.00%
 
 
 ## Usage
@@ -76,19 +76,19 @@ AbstractPhil/gated-david/
 ├── README.md                         # This file
 ├── best_model.json                   # Latest best model info
 ├── weights/
-│   └── david_hierarchical/
-│       └── 20251012_151647/
+│   └── david_balanced/
+│       └── 20251012_152245/
 │           ├── MODEL_SUMMARY.txt     # 🎯 Human-readable performance summary
 │           ├── training_history.json # 📈 Epoch-by-epoch training curve
-│           ├── best_model_acc72.23.safetensors  # ⭐ Accuracy in filename!
-│           ├── best_model_acc72.23_metadata.json
+│           ├── best_model_acc73.29.safetensors  # ⭐ Accuracy in filename!
+│           ├── best_model_acc73.29_metadata.json
 │           ├── final_model.safetensors
 │           ├── checkpoint_epoch_X_accYY.YY.safetensors
 │           ├── david_config.json
 │           └── train_config.json
 └── runs/
-    └── david_hierarchical/
-        └── 20251012_151647/
+    └── david_balanced/
+        └── 20251012_152245/
             └── events.out.tfevents.* # TensorBoard logs
 ```
 
@@ -101,9 +101,9 @@ from huggingface_hub import hf_hub_download
 # Browse available models in MODELS_INDEX.json first!
 
 # Specify model variant and run
-model_name = "david_hierarchical"
-run_id = "20251012_151647"
-accuracy = "72.23"  # From MODELS_INDEX.json
+model_name = "david_balanced"
+run_id = "20251012_152245"
+accuracy = "73.29"  # From MODELS_INDEX.json
 
 # Download config
 config_path = hf_hub_download(
@@ -170,7 +170,7 @@ score = w_anchor * sim(z, anchor) + w_need * sim(z, need) + ...
 ```
 
 ### Fusion Strategy
-**progressive**: Intelligently combines predictions from multiple scales.
+**hierarchical_tree**: Intelligently combines predictions from multiple scales.
 
 ## Training Details
 
@@ -194,7 +194,7 @@ 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: 20251012_151647}
+  note = {Run ID: 20251012_152245}
 }
 ```
 
@@ -209,4 +209,4 @@ Special thanks to Claude (Anthropic) for debugging assistance.
 
 ---
 
-*Generated on 2025-10-12 15:18:32*
+*Generated on 2025-10-12 15:25:16*