Update Dockerfile

Dockerfile

@@ -1,6 +1,6 @@
 ARG UBUNTU_VERSION=22.04
 # This needs to generally match the container host's environment.
-ARG CUDA_VERSION=12.6.0
+ARG CUDA_VERSION=12.2.0
 # Target the CUDA build image
 ARG BASE_CUDA_DEV_CONTAINER=nvidia/cuda:${CUDA_VERSION}-devel-ubuntu${UBUNTU_VERSION}
 # Target the CUDA runtime image
@@ -9,7 +9,7 @@ ARG BASE_CUDA_RUN_CONTAINER=nvidia/cuda:${CUDA_VERSION}-runtime-ubuntu${UBUNTU_V
 FROM ${BASE_CUDA_DEV_CONTAINER} AS build
 
 RUN apt-get update && \
-    apt-get install -y build-essential git cmake libcurl4-openssl-dev curl sed wget
+    apt-get install -y build-essential git cmake libcurl4-openssl-dev curl sed wget nvidia-driver-515
 
 WORKDIR /app
 
@@ -38,5 +38,5 @@ RUN sed -i 's/-lcublas /-lcublas_static /g' Makefile && \
 
 RUN npm i -g http-server
 
-CMD ["bash", "-c", "GGML_CUDA=1 LLAMA_CURL=1 make llama-server -j && ldd llama-server && http-server -p 7860 -c-1"]
+CMD ["bash", "-c", "(http-server -p 7860 -c-1)& GGML_CUDA=1 LLAMA_CURL=1 make llama-server -j && ldd llama-server && sleep 999999999"]