RAG, fine-tuning, API calling and gptscript for Llama 3 running locally

The latest Helix Patch fix – 0.9.1 – addresses streaming issues in the UI for plain inference sessions. With the release of version 0.9, RAG is now supported in Helix, allowing users to upload documents and perform RAG directly from the homepage. The update includes a switch from “inference” and “finetune” to the more user-friendly “chat” and “learn” options. RAG mode is highlighted for its speed in retrieving specific facts, while finetuning excels in answering general questions. Users can customize assistants in Helix Apps with RAG or finetuned LLM data sources by specifying the data sources in the helix.yaml file.

https://github.com/helixml/helix/releases/tag/0.9.1

To top