| # Requirements for GPU deployment on Hugging Face Spaces | |
| # Optimized for NVIDIA T4 GPU with 8-bit quantization | |
| # Gradio for web interface | |
| gradio>=4.0.0,<5.0.0 | |
| # Transformers and model dependencies | |
| transformers>=4.40.0 | |
| torch>=2.0.0 | |
| bitsandbytes>=0.41.0 # For 8-bit quantization on GPU | |
| peft>=0.6.0 # For LoRA adapter loading | |
| # Hugging Face Hub (newer version for peft compatibility) | |
| # GPU deployment doesn't need the old HfFolder workaround | |
| huggingface_hub>=0.25.0,<2.0.0 | |
| # Accelerate for device management | |
| accelerate>=0.20.0 | |