Normalize cosine similarity to [0,1] range

- Vector search now normalizes cosine similarity from [-1,1] to [0,1]
- Cache get also normalized to match
- Fixes threshold comparisons (0.85-0.9 now meaningful)
- Prevents cache miss due to similarity range mismatch

Author: sushanthpy

src/toondb/database.py

@@ -2283,6 +2283,8 @@ def cache_get(
 
                         if query_norm > 0 and cached_norm > 0:
                             score = dot_product / (query_norm * cached_norm)
+                            # Normalize from [-1, 1] to [0, 1] for threshold comparisons
+                            score = (score + 1.0) / 2.0
                         else:
                             score = 0.0
 

src/toondb/namespace.py

@@ -594,6 +594,8 @@ def _vector_search(self, request: SearchRequest) -> SearchResults:
 
             if query_norm > 0 and doc_norm > 0:
                 similarity = dot_product / (query_norm * doc_norm)
+                # Normalize from [-1, 1] to [0, 1] for threshold comparisons
+                similarity = (similarity + 1.0) / 2.0
             else:
                 similarity = 0.0