Starstruk Logo MarkStarstruk

v1.3.7

HKUDS/LightRAGv1.3.7May 14, 2025by danielaskdd

AI Summary

Query parameter improvements and PostgreSQL optimizations including connection configuration and cache persistence.

Key Highlights

  • Added user prompt support to QueryParam
  • Deduplicated merged relation keywords
  • Allowed max_connections to be configured in PostgreSQL
  • Implemented persistent LLM cache on error
  • Optimized query context format for mix and naive mode queries

New Features

  • QueryParam user prompt support
  • PostgreSQL connection configuration
  • LLM cache persistence
  • Query context optimization

Full Release Notes

## What's New
* Feat: Add user prompt to QueryParam by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1546 https://github.com/HKUDS/LightRAG/pull/1552
* Deduplicate merged relation keywords by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1550
* Allow max_connections to be configured in postgres by @maharjun in https://github.com/HKUDS/LightRAG/pull/1547
* Persistent LLM cache on error by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1511
* Use Openai Client Context Manager by @maharjun in https://github.com/HKUDS/LightRAG/pull/1548
* Adding Tongyi OpenAI demo to use Qwen model by @geekben in https://github.com/HKUDS/LightRAG/pull/1518
* Made bedrock_complete generic by @balaji-munusamy in https://github.com/HKUDS/LightRAG/pull/1508
* Optimize query context format for mix and naive mode query by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1545
* Update Dockerfile and docker-compose.yml by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1570
* Update ollama LLM driver and sample code by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1572

## Bugs fixed
* Fix node names with JSON formatting were mistakenly interpreted as JSON objects for PostgreSQL AGE graph storage by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1512
* Fix JSON handling error for PostgreSQL graph storage by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1519
* Update postgres_impl.py syntax to create tables section by @widgit in https://github.com/HKUDS/LightRAG/pull/1524
* Fix time zone and created_at problem  by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1510
* Eliminate redundant chunk data fetching in naive and mix query mode by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1537
* Ignore chat history in vector search by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1538
* Fix streaming problem for OpenAI by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1553
* Resolve json code block freezing issue in WebUI by @danielaskdd in https://github.com/HKUDS/LightRAG/pull/1539
* security fix by @venkateshpabbati in https://github.com/HKUDS/LightRAG/pull/1325

## New Contributors
* @balaji-munusamy made their first contribution in https://github.com/HKUDS/LightRAG/pull/1508
* @geekben made their first contribution in https://github.com/HKUDS/LightRAG/pull/1518
* @maharjun made their first contribution in https://github.com/HKUDS/LightRAG/pull/1547
* @venkateshpabbati made their first contribution in https://github.com/HKUDS/LightRAG/pull/1325

**Full Changelog**: https://github.com/HKUDS/LightRAG/compare/v1.3.6...v1.3.7