Run Analyse on bucket Qwen3-Coder GGUF via ZeroGPU llama.cpp

requirements.txt

@@ -1,10 +1,9 @@
 gradio
 spaces
-torch
-transformers>=4.45.0
-accelerate
-sentencepiece
-einops
+huggingface_hub>=0.30.0
+# CUDA wheel for ZeroGPU (no local nvcc build)
+--extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cu124
+llama-cpp-python
 requests
 gTTS
 pypdf