gardel/lightrag
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add support for list input in quantize_embedding function

Browse Source 
- Convert list to numpy array if needed
- Maintain existing functionality
This commit is contained in:
yangdx
2025-02-01 09:55:05 +08:00
parent 1192727be7
commit 2ba36f87e3
1 changed files with 5 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
lightrag/utils.py
View File

@@ -463,8 +463,12 @@ def cosine_similarity(v1, v2):
return dot_product / (norm1 * norm2) return dot_product / (norm1 * norm2)
def quantize_embedding(embedding: np.ndarray, bits=8) -> tuple: def quantize_embedding(embedding: Union[np.ndarray, list], bits=8) -> tuple:
"""Quantize embedding to specified bits""" """Quantize embedding to specified bits"""
# Convert list to numpy array if needed
if isinstance(embedding, list):
embedding = np.array(embedding)
# Calculate min/max values for reconstruction # Calculate min/max values for reconstruction
min_val = embedding.min() min_val = embedding.min()
max_val = embedding.max() max_val = embedding.max()