docker: CUDA

Dockerfile

@@ -1,11 +1,11 @@
-FROM ghcr.io/ggml-org/llama.cpp:full
+FROM ghcr.io/ggml-org/llama.cpp:full-cuda
 
 # Check if user with UID 1000 exists, if not create it
 RUN id -u 1000 &>/dev/null || useradd -m -u 1000 user
 USER 1000
 
 ENV HOME=/home/user \
-    PATH=${PATH}:/home/user/.local/bin:/app
+    PATH=${PATH}:/home/user/.local/bin:/usr/local/nvidia/bin:/app
 
 WORKDIR ${HOME}/app
 
@@ -14,7 +14,7 @@ COPY --chown=1000 requirements.txt ${HOME}/app
 RUN pip install --no-cache-dir -r requirements.txt
 
 ENV PYTHONPATH=${PYTHONPATH}:${HOME}/.local/bin \
-    LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/app \
+    LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda/lib64:/app \
     HF_HUB_ENABLE_HF_TRANSFER=1 \
     GRADIO_ALLOW_FLAGGING=never \
     GRADIO_NUM_PORTS=1 \