Migrate to Nebius DeepSeek AI with Enhanced Features
#3
by
mohalmah
- opened
Summary
This PR migrates the chat application from HuggingFace Inference API to Nebius AI Studio's DeepSeek-V3 model, providing enhanced AI capabilities and better user experience.
Changes Made
- Replaced HuggingFace InferenceClient with OpenAI-compatible Nebius API
- Integrated DeepSeek-V3 model for superior AI responses
- Updated dependencies: openai>=1.0.0 instead of huggingface_hub
- Enhanced error handling with user-friendly messages
- Improved UI with better default settings and descriptions
- Increased max tokens limit to 4096 for longer responses
- Added comprehensive documentation and setup instructions
- Added API key validation and helpful warnings
Technical Details
- Uses Nebius AI Studio's OpenAI-compatible endpoint
- Requires NEBIUSAISTUDIOAPIKEY environment variable
- Maintains streaming responses for real-time chat experience
- Backward compatible parameter configuration
Testing
- Local testing completed
- Error handling verified
- Streaming responses working
- Parameter controls functional
Ready to merge!
mohalmah
changed pull request status to
merged