Spaces:
Build error
Build error
Commit History
-
Update README with port 7860 configuration
- Modify localhost URLs to use port 7860 consistently - Update Docker run and API documentation endpoints - Align README with recent port configuration changesef37337 -
Add HuggingFace token environment variable loading
- Add comment to clarify HuggingFace token environment variable - Ensure secure and explicit token retrieval from environment - Maintain consistent approach to sensitive configuration managementaf05832 -
Final version with InferenceEndpoint
41bc08e -
Uncomment and activate Uvicorn server configuration
- Restore main block for running FastAPI application - Enable direct server startup with Uvicorn - Set host to 0.0.0.0 and port to 7860 - Simplify application runtime initializationf0a3a20 -
Remove accelerate dependency from requirements.txt
- Streamline project dependencies - Remove unnecessary accelerate library - Maintain minimal and focused requirements configurationbd36812 -
Uncomment and activate LangChain and HuggingFace integration
- Restore full model initialization and generation workflow - Enable HuggingFace InferenceClient with Qwen2.5-72B-Instruct - Reactivate LangGraph-based message processing endpoint - Uncomment environment variable loading and token verification78c6106 -
Uncomment HuggingFace token verification in Dockerfile
- Enable runtime secret verification for HuggingFace token - Remove previous comment blocking token check - Maintain existing Dockerfile security and configuration approach8099c42 -
Remove unused dependencies from requirements.txt
- Remove Clarifai and Pillow libraries - Streamline dependencies to reduce project complexity - Maintain core requirements for language model and application functionality347b9c8 -
Upgrade Python base image to 3.13-slim and adjust Dockerfile
- Update base image from python:3.9 to python:3.13-slim - Change exposed port from commented 8000 to 7860 - Maintain existing Dockerfile structure and user configurationb4b56e1 -
Simplify Dockerfile and app.py configuration
- Remove commented-out HuggingFace and environment-related code - Add user creation in Dockerfile for improved security - Update base image to standard Python 3.9 - Modify file copying to use chown for better permissions - Adjust CMD to use port 7860 consistentlyf49bce4 -
Update Uvicorn server port from 8000 to 7860
- Modify default server port in app.py - Adjust host binding to maintain existing configuration - Ensure compatibility with deployment environmentaf710fb -
Temporarily disable HuggingFace model and generation logic
- Comment out entire model initialization, graph workflow, and generation endpoint - Preserve code structure for potential future reactivation - Remove active model and API call implementations - Keep FastAPI application and basic routing intactd0f6aae -
Switch from local SmolLM2 model to Qwen2.5-72B-Instruct via HuggingFace Inference API
- Replace local model loading with HuggingFace InferenceClient - Update call_model function to use new API-based chat completion method - Simplify model initialization and remove local model loading code - Switch from SmolLM2-135M-Instruct to Qwen2.5-72B-Instruct modele55b892 -
Switch to smaller SmolLM2 model variant
- Change model from SmolLM2-1.7B-Instruct to SmolLM2-135M-Instruct - Update comments to improve clarity and readability - Maintain existing cache directory and authentication configuration543aa6c -
Configure cache directories and add system prompt for local model
- Set specific cache directories for Transformers and HuggingFace libraries - Add system prompt to guide model's language-adaptive behavior - Modify call_model function to include system prompt in message generation - Improve model initialization with explicit cache directory specification4aaa328 -
Switch from HuggingFace InferenceClient to local model loading
- Replace InferenceClient with local model loading using transformers library - Use AutoModelForCausalLM and AutoTokenizer for direct model initialization - Create a text generation pipeline with custom generation parameters - Modify call_model function to work with local model generation - Improve token and model loading logging34aabf7 -
Add accelerate library to requirements for model optimization
- Include accelerate library with minimum version 0.26.0 - Supports potential performance improvements for HuggingFace model inference167addd -
Update Dockerfile to handle HuggingFace token secret and adjust port
- Add Docker secret mounting for HuggingFace token - Verify secret file existence with a simple check - Revert server port back to 8000d655e0b -
Add environment variable loading for HuggingFace token
- Load HuggingFace token from .env file - Pass token to InferenceClient initialization - Update generate endpoint route to remove '/api' prefix - Print token for debugging purposes956e282 -
Refactor app.py with improved model initialization and endpoint structure
- Rename HuggingFace client to 'model' for clarity - Add docstring to call_model function - Update generate endpoint route to '/api/generate' - Remove commented-out test-token endpoint - Enhance code readability and documentatione4413df -
Re-enable HuggingFace model integration and restore generate endpoint
- Uncomment HuggingFace client initialization - Restore model calling and graph configuration - Re-implement generate endpoint with improved docstring - Add detailed error handling for text generatione2f16cb -
Add app_port configuration to README.md metadata
3896f1b -
Update Dockerfile to change FastAPI server port from 8000 to 7860
27abb6b -
Temporarily disable HuggingFace model integration and simplify FastAPI endpoints
- Comment out HuggingFace client initialization - Remove model calling and graph configuration - Simplify root endpoint with more descriptive function name - Comment out generate and test-token endpointsb0424cf -
Remove HuggingFace token validation from Dockerfile
7f9fcdb -
Switch HuggingFace model from Qwen to SmolLM for lightweight inference
a824dd2 -
Refactor HuggingFace model integration and simplify token handling
- Replace HuggingFaceEndpoint with InferenceClient for direct API interaction - Remove environment variable loading and token logging - Add message conversion between LangChain and HuggingFace formats - Implement a new /test-token endpoint for authentication verification - Simplify model invocation and response processing159fcfe -
Add token logging and debug information for HuggingFace authentication
20d2599 -
Modify Hugging Face token validation in Dockerfile
0e25edd -
Add Hugging Face token validation in Dockerfile
9c9fef1 -
Remove unused import from typing module
43b3ac1 -
Update project documentation and dependencies to English, modify HuggingFace model initialization
5f2f118 -
Update pydantic requirement to version 2.0.0 or higher
016d983 -
FIrst commit
f00a406 -
initial commit
Verified signature4e00035