𝗧𝗟;𝗗𝗥:
HuggingChat's default LLM, Cohere's Command R+, can now utilize AI applications built by the community (ZeroGPU Spaces) for:
- Image generation
- Image editing
- Web search
- Document RAG
- And more!
For the release, we've integrated six tools, with plans to expand to more models and tools in the coming weeks. Community members can also add their own tools (any ZeroGPU space is compatible).
For more info, check out the discussion link in the comments 🔥
What's New
The announcement of the beta release of Tools on HuggingChat! Tools expand the capabilities of chat models, allowing them to determine when to use a tool, select the appropriate tool, and pass the necessary arguments (via function calling).
Available Tools
- Web Search: Query the web and retrieve relevant content.
- URL Fetcher: Fetch text content from a specified URL.
- Document Parser: Parse content from PDFs, text files, CSVs, JSONs, and more.
- Image Generation: Create images from text prompts.
- Image Editing: Modify images based on text instructions.
- Calculator: Evaluate mathematical expressions.
How The Tools Where Chosen
- Must be a ZeroGPU Space with exposed API endpoints.
- Fast response time (~25 seconds max).
- Prefer simple, fun tools over complex workflows.
Next Steps
- Enable tool use with previously generated files.
- Allow users to add their own ZeroGPU Spaces as tools.
- Regularly add more official tools.
- Support more models, potentially starting with Llama-3.
- Implement multi-step tool use (Agents).
- Add runtime tools via OpenAPI specification.
For more details and to join the conversation, visit HuggingChat.
Top comments (0)