Update README.md

README.md

@@ -50,10 +50,10 @@ The three models were then distilled into a small ViT model with 11.8 Million Pa
 
 ## Performance
 
-- Validation Set: 94% accuracy
+- Validation Set: 74% accuracy
   - Held out from training data to assess generalization
 
-- Custom Real-World Set: 84% accuracy
+- Custom Real-World Set: 72% accuracy
   - Composed of self-captured images and online-sourced images
   - Designed to be more representative of internet-based images