fastapi
transformers
torch
uvicorn
python-dotenv
optimum[onnxruntime]  # For CPU optimizations with ONNX Runtime
accelerate            # For managing multi-device setup (CPU/GPU)
gunicorn              # For running multiple workers