# Requirements for Hugging Face Space deployment # Optimized for CPU inference with llama.cpp (GGUF format) # Gradio for web interface # Use stable 4.x version gradio>=4.0.0,<5.0.0 # huggingface_hub: Pin to version with HfFolder (required by Gradio) # HfFolder was removed in huggingface_hub >= 0.23.0 huggingface_hub>=0.20.0,<0.23.0 # llama-cpp-python: Use pre-built wheel (includes Qwen3 support) llama-cpp-python==0.3.16