🇷🤝💬 R-help-chat

Chat with the R-help mailing list archives. An LLM turns your question into a search query, including year ranges and months, and generates an answer from the retrieved emails. You can ask follow-up questions with the chat history as context. ➡️ To clear the history and start a new chat, press the 🗑️ clear button. Answers may be incorrect.
Compute Mode

NOTE: remote mode does not use ZeroGPU

🌐 Now in remote mode, using the OpenAI API
⚠️ Privacy Notice: Data sharing with OpenAI is enabled
✨ text-embedding-3-small and gpt-4o-mini
🏠 See the project's GitHub repository

Database: 51566 emails from January 2015 to August 2025. Features: RAG, today's date, hybrid search (dense+sparse), multiple retrievals, citations output (remote), chat memory. Tech: LangChain + Hugging Face + Gradio; ChromaDB and BM25S-based retrievers.
Click an example to fill the message box
Multiple retrievals
Asking follow-up questions