20270722

What's Changed

Precision Improvement

• How to choose chunksize and splitter
• Hybrid knowledge graph and dense retrieval to improve precision
• Improve SFT annotation based on the scaling law
• Support llm reranker

LLM API

• Support silicon cloud and internlm remote API
• Add magic prompt for kimi

Feature

• Wechat support wkteam
• Learning through the chat group
• Support pyppeteer
• InternLM Summer Camp3 by @boshallen in #321

New Contributors

• @eltociear made their first contribution in #274
• @boshallen made their first contribution in #321

Full Changelog: 2024041...2024072

20240415

What's Changed

Frontend

• Add web server and front-end by @JiaYingLii @Echo-minn @lanbitou-tech @fly2tomato
• Implement Android APP support WeChat group by @lmhh
• Support Lark @tpoisonooo

Feature

• Support rag.py for SFT data annotation @tpoisonooo
• Support deepseek, xi-api, alles-apin, puyu and kimi auto model selection @tpoisonooo
• Update architecture_zh.md by @tpoisonooo in #63
• Add for 2G memory by @tpoisonooo in #100
• Support pdf,word, excel, pptx and html file format by @tpoisonooo
• Add github issue crawler by @sanbuphy in #69
• Support zhipuai and add test query gradio by @sanbuphy in #116
• Split feature_store and retriever by @tpoisonooo in #99

Docs

• Add WeChat and architecture by @JimmyMa99 @tpoisonooo
• Add FAQ by @Modas-Li in #44
• Add arch by @tpoisonooo in #61
• Fix No module named 'faiss.swigfaiss_avx2' by @chg0901 in #215
• New clean REAMDE @tpoisonooo

Fix

• Fix installation by @LzMingYueShanPao in #224
• De-serialization relies loading a pickle file need allow_dangerous_deserialization=True by @weedge in #151
• Fix installation by @chg0901 in #217
• Change sg_search prompt KEYWORDS_TEMPLATE, between keywords use , separate by @weedge in #158
• Use List instead of list by @LzMingYueShanPao in #222

New Contributors

• @JimmyMa99 made their first contribution in #18
• @Modas-Li made their first contribution in #44
• @fly2tomato made their first contribution in #46
• @JiaYingLii made their first contribution in #54
• @sanbuphy made their first contribution in #69
• @lmhh made their first contribution in #47
• @weedge made their first contribution in #151
• @chg0901 made their first contribution in #215
• @LzMingYueShanPao made their first contribution in #222

Full Changelog: v0.1.0rc1...2024041

Init project

HuixiangDou is a group chat assistant based on LLM (Large Language Model).

Advantages:

1. Design a two-stage pipeline of rejection and response to cope with group chat scenario, answer user questions without message flooding, see arxiv2401.08772
2. Low cost, requiring only 1.5GB memory and no need for training
3. Offers a complete suite of Web, Android, and pipeline source code, which is industrial-grade and commercially viable

Check out the scenes in which HuixiangDou are running and join WeChat Group to try AI assistant inside.

If this helps you, please give it a star ⭐