feat: Better cublas support in vLLM container. (#4514)

Signed-off-by: jthomson04 <[email protected]>

Author: jthomson04

container/Dockerfile.vllm

@@ -217,8 +217,8 @@ COPY --from=framework /usr/local/cuda/bin/fatbinary /usr/local/cuda/bin/fatbinar
 COPY --from=framework /usr/local/cuda/include/ /usr/local/cuda/include/
 COPY --from=framework /usr/local/cuda/nvvm /usr/local/cuda/nvvm
 COPY --from=framework /usr/local/cuda/lib64/libcudart.so* /usr/local/cuda/lib64/
-COPY --from=framework /usr/local/cuda/lib64/libcublas.so* /usr/local/cuda/lib64/
-COPY --from=framework /usr/local/cuda/lib64/libcublasLt.so* /usr/local/cuda/lib64/
+RUN ln -s /usr/local/cuda/lib64/libcublas.so.12 /usr/local/cuda/lib64/libcublas.so
+RUN ln -s /usr/local/cuda/lib64/libcublasLt.so.12 /usr/local/cuda/lib64/libcublasLt.so
 
 ### COPY NATS & ETCD ###
 # Copy nats and etcd from dev image