Merged
Conversation
Implements multi-granular embeddings (entity_dense and relation_dense) for improved code retrieval, including two-stage prefetch search in Qdrant, fusion logic in ranking, and ingestion pipeline changes to generate and store new vectors. Updates configuration and collection management to support the new vector types, enabling more precise and graph-aware retrieval as outlined in the project plan.
Introduces adaptive weighting and fusion for entity and relation dense vectors in hybrid search, based on query characteristics. Updates ranking logic to detect graph-oriented queries and adjust weights accordingly. Enables multi-granular vector search and fusion in hybrid_search.py, and aligns relation dense vector dimension with embedding model output in config.py.
🤖 Augment PR SummarySummary: Adds optional multi-granular retrieval (entity + relation embeddings) and improves graph/CLI ergonomics to better handle symbol- and dependency-oriented queries. Changes:
Technical Notes: Multi-granular querying falls back to standard dense search if prefetch isn’t supported (older Qdrant) or vectors aren’t available; ingest gates vector writes on the collection’s advertised vector names to avoid upsert errors. 🤖 Was this summary useful? React with 👍 or 👎 |
![augmentcode[bot]](https://avatars.githubusercontent.com/in/1027498?s=60&v=4){kind=small}
Introduced get_qdrant_url_for_host to ensure the CLI uses localhost instead of Docker's internal hostname for Qdrant connections. Updated graph and reset commands to use this helper. Enhanced the reset command to clear local and dev-workspace cache files and symbols directories before running the indexer.
The reset command now stops services and removes volumes with 'docker-compose down -v --remove-orphans'. The indexer is started in detached mode to avoid blocking the CLI, and users are informed how to monitor its logs.
Replaces 'docker compose ps' with 'docker ps' using label filters to allow status checks from any directory, not just where docker-compose.yml is present. Adapts output parsing to maintain compatibility with previous format.
Refactors hybrid_search.py to apply RRF fusion per-query for multi-granular vectors, ensuring correct ranking semantics. Updates pipeline.py to only enable multi-granular vector indexing if the collection supports the required vectors, and adds logic to batch embed and store entity and relation vectors during ingestion.
Collaborator
Author
|
augment review |
Updated Docker service status check to use Go template JSON output for broader compatibility and improved parsing robustness. In hybrid_search.py, adjusted multi_granular_query call to ignore unused mg_results, clarifying intent for RRF fusion.
Update docker-compose.neo4j.yml for Neo4j 5.x memory settings. Enhance backend to set Symbol node properties on creation and update, and add a simple degree-based pagerank computation without requiring GDS. Use COALESCE in Cypher queries to avoid warnings for missing properties. Update CLI status command to only show running containers.
Collaborator
Author
|
augment review |
Refines score accumulation in hybrid ranking to sum entity and relation scores for multi-query scenarios, ensuring accurate contribution tracking. Updates Docker Compose project name handling in status command for better compatibility with custom project names. Enhances Cypher query in Neo4j backend to prefer more specific start_line values and improves language field merging. Adds entity_dense and relation_dense fields to hybrid search result components for better transparency.
Collaborator
Author
|
augment review |
Changed the IMPORTS upsert logic to prefer the incoming language value for consistency with CALLS upsert. Updated the reset CLI command to forcibly remove any stale indexer container before running a new one with --rm for idempotency.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.