Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

GitHub - alexpinel/Dot: Standalone app for fully local RAG with Mistral 7B

Apr 06, 2024 - github.com
The article introduces Dot, a standalone open source app designed for easy use of local LLMs and Retrieval Augmented Generation (RAG) to interact with documents and files. The app is completely standalone and comes with all dependencies, including a copy of Mistral 7B, making it accessible to users without any prior knowledge of programming or local LLMs. Dot supports multiple document formats and allows users to load multiple documents into an LLM and interact with them in a fully local environment.

Dot is built with Electron JS and comes with a bundled install of Python that contains all necessary libraries and files. The app can be installed from the Dot website or set up for development. Future features to be added include Linux support, ability to choose LLM, image support, increased document awareness, loading individual files, increased security considerations, support for more docs, and storing file databases. The author encourages contributions from others as they are a busy student working on this as a side project.

Key takeaways:

  • Dot is a standalone open source app designed for easy use of local LLMs and RAG to interact with documents and files, with no prior knowledge of programming or local LLMs required.
  • Dot can load multiple documents into an llm and interact with them in a fully local environment through Retrieval Augmented Generation (RAG), supporting documents such as pdf, docx, pptx, xlsx, and markdown.
  • Dot is built with Electron JS, and its main functionalities come from a bundled install of Python that contains all libraries and necessary files, including llama.cpp, FAISS, Langchain & Huggingface.
  • The author plans to add future features to Dot, including Linux support, the ability to choose LLM, image support, increased security considerations, support for more docs, and storing file databases.
View Full Article

Comments (0)

Be the first to comment!