gmakstutis/LightRAG
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
5284 commits 5 branches 72 tags 705 MiB
Commit graph

3386 commits

Author SHA1 Message Date
yangdx
99e58ac752 fix: add retry mechanism for Memgraph transient errors 
- Implement exponential backoff retry for transaction conflicts
- Add tenacity-based retry decorator with 5 attempts
- Handle TransientError in upsert_node and upsert_edge operations
- Resolve "Cannot resolve conflicting transactions" errors
- Improve system reliability under concurrent load
 2025-07-19 10:34:35 +08:00
yangdx
aea18d7b80 Merge branch 'fix-redis' into fix-keyed-lock 2025-07-19 10:26:08 +08:00
yangdx
6e1657a771 Improve thread safety for relationship rebuilding 
- Sort src and tgt for consistent lock keys
- Maintain order-independent locking
 2025-07-19 10:25:48 +08:00
yangdx
96b94acc83 Enhance Redis connection handling with retries and timeouts 
- Added Redis connection timeout configurations
- Implemented retry logic for Redis operations
- Updated error handling for timeout cases
- Improved connection pool management
- Added environment variable support
 2025-07-19 10:15:26 +08:00
yangdx
05bc5cfb64 Improve task execution with early failure detection 
- Add early failure detection for async tasks
- Cancel pending tasks on first exception
 2025-07-19 10:14:22 +08:00
yangdx
12d4f12e57 fix: sort edge_key components in _locked_process_edges for consistent locking 
- Ensures bidirectional relationships use same lock key
- Maintains thread safety for knowledge graph edge operations
 2025-07-19 07:36:50 +08:00
Daniel.y
801bdc520e
Merge pull request #1801 from Chida82/main 
fix Azure deployment
 2025-07-18 16:00:09 +08:00
yangdx
f033fd6f87 fix(postgres): improve AGE agtype parsing and simplify error logging 
- Fix JSON parsing errors caused by :: characters in data content
- Implement precise agtype string parsing using rfind() to separate JSON content from type identifiers
- Add robust error handling for malformed JSON in graph data
 2025-07-18 08:50:47 +08:00
Dario Chini
5b28233903 fix Azure deployment 2025-07-17 23:11:07 +02:00
yangdx
488028b9e2 Remove separate requirements.txt and update Dockerfile to use pip install 2025-07-18 01:58:46 +08:00
yangdx
99527027de feat: change default query mode from hybrid to mix 
- Update default mode for Ollama chat endpoint
- Update default mode for query endpoint of LightRAG
 2025-07-17 19:21:15 +08:00
yangdx
e828539b24 Update README 2025-07-17 19:05:34 +08:00
yangdx
be2d938c84 Fix file path handling in graph operations 
- Filter out empty file paths
- Handle missing file_path fields
 2025-07-17 18:33:14 +08:00
yangdx
b321afefaa Bump core version to 1.4.3 and api version to 0186 2025-07-17 16:58:57 +08:00
yangdx
f3c0dab7ce Bump core version to 1.4.2 and api version to 0185 2025-07-17 12:26:10 +08:00
yangdx
14d9fe49b0 refactor(milvus): remove entity_type and weight fields from schema 
- Remove entity_type field from entities collections
- Remove weight field from relationships collections
- Update schema definitions and index creation logic
- Maintain backward compatibility with existing data via dynamic fields
 2025-07-17 12:08:35 +08:00
yangdx
7184c7b3ab fix: change default edge weight from 0.0 to 1.0 in entity extraction and graph storage 
- Update extract_entities function in operate.py to use 1.0 as default weight
- Fix Neo4j implementation to use 1.0 instead of 0.0 for missing edge weights
- Fix Memgraph implementation to use 1.0 instead of 0.0 for missing edge weights
- Ensures consistent non-zero default weights across all graph storage backends
 2025-07-17 11:30:49 +08:00
yangdx
57c8c19628 Add datetime format migration for doc status table 2025-07-16 22:21:51 +08:00
yangdx
910c6973f3 Limit file deletion to current directory only after document cleaning 2025-07-16 20:35:24 +08:00
yangdx
c7b566f6d5 Fix cache migration MD5 error for PostgreSQL 2025-07-16 19:24:57 +08:00
yangdx
80f7e37168 Fix default workspace name for PostgreSQL AGE graph storage 2025-07-16 19:16:22 +08:00
yangdx
bab2803953 Optimize PostgreSQL database migrations for LLM cache 
- Combine column migration into single operation
- Optimize LLM cache key migration query
- Improve migration error handling
- Add conflict detection for cache migration
 2025-07-16 17:32:53 +08:00
yangdx
bd340fece6 Fix timestamp column migration comment typos 
- Correct timezone-related comments
- Fix typo in debug log message
- Update migration success message
- Maintain same migration logic
 2025-07-16 14:27:52 +08:00
yangdx
2bf0d397ed Update webui assets 2025-07-16 10:18:51 +08:00
yangdx
e4f62de727 Bump aip version to 0184 2025-07-16 04:57:46 +08:00
yangdx
500e940f75 Remove max token summary display from splash screen 2025-07-16 04:55:32 +08:00
yangdx
8103b200db Set DEFAULT_HISTORY_TURNS to 0 2025-07-16 02:20:27 +08:00
yangdx
0adb5f2595 Update webui assets 2025-07-16 01:39:48 +08:00
yangdx
45d38fa083 Fix JSON error logging in Redis storage implementations 2025-07-16 01:35:07 +08:00
Daniel.y
b44c8d46a5
Merge pull request #1782 from HKUDS/rerank 
Refactor the token control system
 2025-07-16 00:23:25 +08:00
yangdx
6e084bfae1 Increase default related chunk number from 5 to 10 2025-07-16 00:22:34 +08:00
yangdx
b1276a079f Fix linting 2025-07-15 23:57:24 +08:00
yangdx
5f7cb437e8 Centralize query parameters into LightRAG class 
This commit refactors query parameter management by consolidating settings like `top_k`, token limits, and thresholds into the `LightRAG` class, and consistently sourcing parameters from a single location.
 2025-07-15 23:56:49 +08:00
zrguo
3ead0489b8 Remove "rank", "weight", "keywords" 2025-07-15 21:47:33 +08:00
zrguo
1541034816 Add DEFAULT_RELATED_CHUNK_NUMBER 2025-07-15 21:35:12 +08:00
yangdx
089346f8df Bump api version to 0183 2025-07-15 19:52:50 +08:00
zrguo
42f1fd60f4 Update operate.py 2025-07-15 18:59:52 +08:00
zrguo
29e82723e6 Update operate.py 2025-07-15 18:57:57 +08:00
yangdx
93b25a65d5 Update webui assets 2025-07-15 18:10:00 +08:00
yangdx
bdcb280731 Merge branch 'main' into rerank 2025-07-15 18:09:01 +08:00
yangdx
6d66cde4ac Reorder query settings in web UI 2025-07-15 18:06:00 +08:00
zrguo
e254c3dd81 Update openai.py 2025-07-15 17:30:30 +08:00
yangdx
661a41f9eb Update webui assets 2025-07-15 17:25:39 +08:00
yangdx
1927cb2685 Fix linting 2025-07-15 17:24:57 +08:00
yangdx
47341d3a71 Merge branch 'main' into rerank 2025-07-15 16:12:33 +08:00
yangdx
e8e1f6ab56 feat: centralize environment variable defaults in constants.py 2025-07-15 16:11:50 +08:00
Daniel.y
6d1260aafa
Merge pull request #1766 from HKUDS/fix-memgraph-max-nodes-issue 
Fix Memgraph get_knowledge_graph issues
 2025-07-15 16:07:04 +08:00
zrguo
91d0f65476 Update QueryParam 2025-07-15 14:21:58 +08:00
yangdx
3da9f8aab4 Fix logging output condition in shared_storage.py. Early return if logging disabled 2025-07-15 13:38:05 +08:00
yangdx
ccc2a20071 feat: remove deprecated MAX_TOKEN_SUMMARY parameter to prevent LLM output truncation 
- Remove MAX_TOKEN_SUMMARY parameter and related configurations
- Eliminate forced token-based truncation in entity/relationship descriptions
- Switch to fragment-count based summarization logic using FORCE_LLM_SUMMARY_ON_MERGE
- Update FORCE_LLM_SUMMARY_ON_MERGE default from 6 to 4 for better summarization
- Clean up documentation, environment examples, and API display code
- Preserve backward compatibility by graceful parameter removal

This change resolves issues where LLMs were forcibly truncating entity relationship
descriptions mid-sentence, leading to incomplete and potentially inaccurate knowledge
graph content. The new approach allows LLMs to generate complete descriptions while
still providing summarization when multiple fragments need to be merged.

Breaking Change: None - parameter removal is backward compatible
Fixes: Entity relationship description truncation issues
 2025-07-15 12:26:33 +08:00