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

GitHub - psychic-api/rag-stack: 🤖 Deploy a private ChatGPT alternative hosted within your VPC. 🔮 Connect it to your organization's knowledge base and use it as a corporate oracle. Supports open-source LLMs like Llama 2, Falcon, and GPT4All.

Jul 20, 2023 - github.com
RAGstack is a private ChatGPT alternative that can be hosted within your VPC. It can be connected to an organization's knowledge base and used as a corporate oracle. It supports open-source Large Language Models (LLMs) like Llama 2, Falcon, and GPT4All. The Retrieval Augmented Generation (RAG) technique is used to augment the capabilities of an LLM by retrieving information from other systems and inserting them into the LLM’s context window via a prompt. This provides LLMs with information beyond their training data, which is crucial for most enterprise use cases.

RAGstack can be run locally or deployed to Google Cloud. It deploys resources for retrieval-augmented generation including open-source LLMs, a vector database, and a server + UI. It supports GPT4All, Falcon-7b, and Llama 2 models, and uses Qdrant, an open-source vector database. The server and UI handle PDF uploads, allowing users to chat over their PDFs using Qdrant and their chosen open-source LLM. The roadmap includes support for Llama-2-40b and deployment on AWS.

Key takeaways:

  • RAGstack is a private ChatGPT alternative that can be hosted within a VPC and connected to an organization's knowledge base, supporting open-source LLMs like Llama 2, Falcon, and GPT4All.
  • The RAG (Retrieval Augmented Generation) technique augments the capabilities of a large language model by retrieving information from other systems and inserting them into the LLM’s context window via a prompt.
  • RAGstack can be run locally or deployed to Google Cloud, and it includes resources for retrieval-augmented generation such as open-source LLM, vector database, and a server with UI.
  • The roadmap for RAGstack includes support for Llama-2-40b and deployment on AWS, in addition to the already supported GPT4all, Falcon-7b, and deployment on GCP.
View Full Article

Comments (0)

Be the first to comment!