Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[hotfix] Quantization patch; fix semantic_search_faiss/semantic_search_usearch rescoring #2558

Merged
merged 2 commits into from
Mar 26, 2024
Merged

[hotfix] Quantization patch; fix semantic_search_faiss/semantic_search_usearch rescoring #2558

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
84c075b
Correctly update the scores after rescoring
tomaarsen Mar 25, 2024
72b3e28
Sort scores from high to low in recommended example
tomaarsen Mar 25, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion examples/applications/embedding-quantization/semantic_search_recommended.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -94,7 +94,7 @@ def search(query, top_k: int = 10, rescore_multiplier: int = 4):

# 6. Sort the scores and return the top_k
start_time = time.time()
indices = scores.argsort()[:top_k]
indices = (-scores).argsort()[:top_k]
top_k_indices = binary_ids[indices]
top_k_scores = scores[indices]
sort_time = time.time() - start_time
Expand Down
4 changes: 2 additions & 2 deletions sentence_transformers/quantization.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -140,7 +140,7 @@ def semantic_search_faiss(
rescored_scores = np.einsum("ij,ikj->ik", rescore_embeddings, top_k_embeddings)
rescored_indices = np.argsort(-rescored_scores)[:, :top_k]
indices = indices[np.arange(len(query_embeddings))[:, None], rescored_indices]
scores = rescored_scores[:, :top_k]
scores = rescored_scores[np.arange(len(query_embeddings))[:, None], rescored_indices]

delta_t = time.time() - start_t

Expand Down Expand Up @@ -293,7 +293,7 @@ def semantic_search_usearch(
rescored_scores = np.einsum("ij,ikj->ik", rescore_embeddings, top_k_embeddings)
rescored_indices = np.argsort(-rescored_scores)[:, :top_k]
indices = indices[np.arange(len(query_embeddings))[:, None], rescored_indices]
scores = rescored_scores[:, :top_k]
scores = rescored_scores[np.arange(len(query_embeddings))[:, None], rescored_indices]

delta_t = time.time() - start_t

Expand Down
Loading