v0.23.0: [Docker] Fix CUTLASS DSL cu13 install order in Dockerfile (#45204)
vLLM has released version 0.23.0, with a release candidate v0.23.0rc2 preceding it. Both releases address an issue with the installation order of the CUTLASS DSL for CUDA 13 within Dockerfiles. This fix was originally cherry-picked from a previous commit. AI
IMPACT Minor update to an open-source library for LLM inference, primarily addressing a build issue.