Starstruk Logo MarkStarstruk

v0.5.5

eosphoros-ai/DB-GPTv0.5.5Apr 26, 2024by github-actions[bot]

AI Summary

This release significantly expands database and model support, adding TuGraph and OceanBase databases, along with new LLM support for Qwen2MoE, Llama-3, and CodeQwen1.5. It also introduces advanced retrieval capabilities like CrossEncoder reranking and MRR/HitRate metrics.

Key Highlights

  • Added support for TuGraph graph database
  • Support Qwen2MoE and Llama-3 models
  • Add CrossEncoder rerank for improved relevance
  • Add MRR and HitRate retriever metrics

New Features

  • Added support for TuGraph graph database
  • Add MRR and HitRate retriever metrics
  • Add CrossEncoder rerank
  • Support Qwen2MoE
  • Support Llama-3
  • Support OceanBase vector database
  • Support CodeQwen1.5-7B-Chat
  • Support retry for 'Chat Data'
  • Dbgpts support agent
  • Support oceanbase monaco plugin

Full Release Notes

## ✨ Enhancements

- Added support for TuGraph graph database (#1451)
- Add MRR and HitRate retriever metrics. (#1456)
- Add CrossEncoder rerank (#1442)
- Support Qwen2MoE (#1439)
- Support Llama-3 (#1436)
- Support OceanBase vector database (#1435)
- Datasource support summary refresh (#1433)
- Upgrade pydantic to 2.x (#1428)
- Support CodeQwen1.5-7B-Chat (#1431)
- Support retry for 'Chat Data' (#1419)
- Dbgpts support agent (#1417)
- Support oceanbase monaco plugin (#1386)


## 🐞 Bug fixes

- Fix parameter type mismatch issue when using Milvus (#1449)
- Fix oceanbase vector\_name\_exists bug (#1465)
- Fix clickhouse occur error  AttributeError: 'ClickhouseConnect… (#1432)
- Fix awel check for empty DataFrame data bug (#1430)
- Fix rdbms\_db\_summary (#1409)

Thank you to all our contributors for making this release possible!
@Aries-ckt, @GITHUBear, @HSunboy, @KingSkyLi, @dusens, @fangyinc, @jsRuner and @yyhhyyyyyy