Fix LLAMA_CUBLAS deprecated error: Use GGML_CUDA flag with CPU fallback

Dockerfile

@@ -25,9 +25,13 @@ COPY requirements.txt .
 # Install PyTorch with CUDA support
 RUN pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
 
-# Install llama-cpp-python with CUDA support
-ENV CMAKE_ARGS="-DLLAMA_CUBLAS=on"
-RUN pip install llama-cpp-python --force-reinstall --no-cache-dir
+# Install llama-cpp-python with CUDA support (using new GGML_CUDA flag)
+ENV CMAKE_ARGS="-DGGML_CUDA=on"
+ENV FORCE_CMAKE=1
+# Try CUDA build first, fallback to CPU if it fails
+RUN pip install llama-cpp-python --force-reinstall --no-cache-dir || \
+    (echo "CUDA build failed, installing CPU version..." && \
+     CMAKE_ARGS="" pip install llama-cpp-python --force-reinstall --no-cache-dir)
 
 # Install other requirements
 RUN pip install -r requirements.txt
@@ -44,4 +48,4 @@ ENV HF_HOME=/.cache
 EXPOSE 7860
 
 # Run the application
-CMD ["python", "app.py"]
+CMD ["python", "app.py"]