Update README - Run 20251012_194945

README.md

@@ -21,7 +21,7 @@ model-index:
       type: imagenet-1k
     metrics:
     - type: accuracy
-      value: 74.97
+      value: 75.31
 ---
 
 # David: Multi-Scale Feature Classifier
@@ -55,13 +55,14 @@ exist simultaneously in the same shared space with the correct checks and spacin
 ## Performance
 
 ### Best Results
-- **Validation Accuracy**: 74.97%
-- **Best Epoch**: 4
-- **Final Train Accuracy**: 78.98%
+- **Validation Accuracy**: 75.31%
+- **Best Epoch**: 5
+- **Final Train Accuracy**: 80.20%
 
 ### Per-Scale Performance
-- **Scale 256**: 73.22%
-- **Scale 512**: 74.80%
+- **Scale 256**: 73.62%
+- **Scale 512**: 75.21%
+- **Scale 768**: 74.39%
 
 
 ## Usage
@@ -82,8 +83,8 @@ AbstractPhil/david-shared-space/
 │       └── 20251012_194945/
 │           ├── MODEL_SUMMARY.txt     # 🎯 Human-readable performance summary
 │           ├── training_history.json # 📈 Epoch-by-epoch training curve
-│           ├── best_model_acc74.97.safetensors  # ⭐ Accuracy in filename!
-│           ├── best_model_acc74.97_metadata.json
+│           ├── best_model_acc75.31.safetensors  # ⭐ Accuracy in filename!
+│           ├── best_model_acc75.31_metadata.json
 │           ├── final_model.safetensors
 │           ├── checkpoint_epoch_X_accYY.YY.safetensors
 │           ├── david_config.json
@@ -105,7 +106,7 @@ from huggingface_hub import hf_hub_download
 # Specify model variant and run
 model_name = "david_balanced"
 run_id = "20251012_194945"
-accuracy = "74.97"  # From MODELS_INDEX.json
+accuracy = "75.31"  # From MODELS_INDEX.json
 
 # Download config
 config_path = hf_hub_download(
@@ -214,4 +215,4 @@ Special thanks to Claude (Anthropic) for debugging assistance.
 
 ---
 
-*Generated on 2025-10-12 20:05:57*
+*Generated on 2025-10-12 20:09:05*