Update README.md

README.md

@@ -427,8 +427,7 @@ The temperature and any learned transformations are trained alongside the classi
 
 5. **The 5×5×5 grid scales**: The same sectorization works for both close-up tabletop scenes and panoramic landscapes by leveraging perspective scaling of sector volumes.
 
-6. **Burn test analogy**: Just as 44 images × 227 repeats teaches the geo_prior to anchor on tags like `1person`, synthetic scenes with known spatial structure should teach it to anchor on spatial configurations — but with ground truth labels instead of statistical inference.
-
+6. 
 ---
 
 ## 8. Technical Notes