LangChain-Chatchat
An open-source local knowledge-base question-answering application built on LangChain. It supports a RAG pipeline covering document loading, text splitting, vector-based retrieval and dialogue with local large language models.
Timeline (last 90 days)
LangChain-Chatchat uses the recursive character splitter by default and supports setting the chunk_size and chunk_overlap parameters.
In LangChain-Chatchat's model_settings.yaml, deepseek-r1 and bge-m3 must be explicitly added to the LLM_MODELS and EMBEDDING_MODELS lists, otherwise a model-not-found prompt will appear when initializing the knowledge base.
httpx package version 0.28.1 does not support proxies, which causes the API to fail to return correct JSON when LangChain-Chatchat starts, and it needs to be downgraded to version 0.27.2.
After startup, the LangChain-Chatchat application runs on 127.0.0.1:8501 by default.
This solution also runs in a Windows 11, Python 3.10, CPU environment; CUDA is not mandatory, but the official recommendation is version 11.7 or above.
All Facts (5)
LangChain-Chatchat uses the recursive character splitter by default and supports setting the chunk_size and chunk_overlap parameters.
50%UnverifiedIn LangChain-Chatchat's model_settings.yaml, deepseek-r1 and bge-m3 must be explicitly added to the LLM_MODELS and EMBEDDING_MODELS lists, otherwise a model-not-found prompt will appear when initializing the knowledge base.
50%Unverifiedhttpx package version 0.28.1 does not support proxies, which causes the API to fail to return correct JSON when LangChain-Chatchat starts, and it needs to be downgraded to version 0.27.2.
50%UnverifiedAfter startup, the LangChain-Chatchat application runs on 127.0.0.1:8501 by default.
50%UnverifiedThis solution also runs in a Windows 11, Python 3.10, CPU environment; CUDA is not mandatory, but the official recommendation is version 11.7 or above.
50%