AnythingLLM adds Ollama: fully local document chat
In one sentence On December 27, 2023 AnythingLLM, Mintplex Labs' open source (MIT) app for chatting with your own documents, adds Ollama as a model provider. With LocalAI, LM Studio, the native embedder and built-in LanceDB, the RAG pipeline can run without cloud services.
AnythingLLM is an open source web application by Mintplex Labs, published on GitHub in June 2023: you upload documents, organize them into separate "workspaces" and ask questions to a language model, which answers with citations to the sources (a technique called RAG).
On December 27, 2023 the project adds Ollama to its model providers. At that point the list includes both commercial services (OpenAI, Azure OpenAI, Anthropic, Google Gemini) and local options: llama.cpp-compatible models, Ollama, LM Studio and LocalAI.
The rest of the chain can stay in-house too: the default embedder is AnythingLLM's native one and the default vector database is LanceDB, bundled with the application. Anyone who wants a private knowledge base over internal documents can run it on their own PC or server.
Companies
Mintplex Labs
Tools
AnythingLLM, Ollama, LocalAI, LM Studio, LanceDB
Tags
Sources