Locallab

Added
- Added comprehensive API documentation display on server startup with curl examples
- Added model configuration section that displays current model and optimization settings
- Added system instructions section showing the current prompt template
- Improved environment variable handling for model configuration
- Enhanced server startup logging with detailed optimization settings
- Added support for reading HUGGINGFACE_MODEL environment variable to specify model
- Redesigned modern ASCII art banners for a more aesthetic interface
- Improved UI with cleaner banner separations and better readability

Fixed
- Fixed parameter mismatch in text generation endpoints by properly handling `max_new_tokens` parameter
- Resolved coroutine awaiting issues in streaming generation endpoints
- Fixed async generator handling in `stream_chat` and `generate_stream` functions
- Enhanced error handling in streaming responses to provide better error messages
- Improved compatibility between route parameters and model manager methods

Added
- Added missing dependencies in `setup.py`: huggingface_hub, pynvml, and typing_extensions
- Improved dependency management with dev extras for testing packages
- Enhanced error handling for GPU memory detection

Fixed
- Fixed circular import issues between modules
- Improved error handling in system utilities
- Enhanced compatibility with Google Colab environments

Added
- New model loading endpoint that accepts model_id in the request body at `/models/load`
- `format_chat_messages` function to properly format chat messages for the model
- CLI function to support command-line usage with click interface

Fixed
- Properly awaiting async `generate_text` in chat completion endpoint
- Fixed async generator handling in `generate_stream` function
- Fixed streaming in the `stream_chat` function to correctly send server-sent events
- Properly escaped newline characters in the streaming response
- Added missing dependencies in `setup.py`: colorama, python-multipart, websockets, psutil, and nest-asyncio

Added
- `get_network_interfaces` function to retrieve information about available network interfaces
- `get_public_ip` async function to retrieve the public IP address of the machine
- Adapter methods in `ModelManager` (`generate_text` and `generate_stream`) to maintain API compatibility with route handlers

Fixed
- Import error for `get_public_ip` and `get_network_interfaces` functions
- Naming mismatch between route handlers and `ModelManager` methods
- New dependencies in `setup.py`: `netifaces` and `httpx`

Fixed
- Fixed API endpoint errors for `/models/available` and other model endpoints
- Resolved parameter error in `get_model_generation_params()` function
- Improved error handling for model optimization settings through environment variables
- Fixed circular import issues between routes and core modules
- Enhanced Flash Attention warning message to be more informative

Added
- Added new `get_gpu_info()` function for detailed GPU monitoring
- Added improved system resource endpoint with detailed GPU metrics
- Added robust environment variable handling for optimization settings

Changed
- Made optimization flags more robust by checking for empty string values
- Improved fallback handling for missing torch packages
- Enhanced server startup logs with better optimization information