danvics/haiku.rag
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Improve performance of vector search score normalization

Browse source 
Co-authored-by: Tres Seaver <tseaver@palladion.com>
This commit is contained in:
Yiorgis Gozadinos 2025-11-04 16:34:47 +02:00
parent 5ffedf013b
commit 4773b56f9d
No known key found for this signature in database
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
src/haiku/rag/store/repositories/chunk.py
View file

@ -374,7 +374,7 @@ class ChunkRepository:
"""Extract scores from DataFrame columns based on search type.""" """Extract scores from DataFrame columns based on search type."""
if "_distance" in df.columns: if "_distance" in df.columns:
# Vector search - convert distance to similarity # Vector search - convert distance to similarity
return [max(0.0, 1.0 / (1.0 + d)) for d in df["_distance"].tolist()] return ((df["_distance"] + 1).rdiv(1)).clip(lower=0.0).tolist()
elif "_relevance_score" in df.columns: elif "_relevance_score" in df.columns:
# Hybrid search - relevance score (higher is better) # Hybrid search - relevance score (higher is better)
return df["_relevance_score"].tolist() return df["_relevance_score"].tolist()