Switch to release branch

Author: Apsu

.github/workflows/build-vllm.yaml

@@ -16,7 +16,6 @@ env:
   XFORMERS_BUILD_VERSION: 0.0.29.post2+cu124
   FLASHINFER_REF: v0.2.2.post1
   FLASHINFER_BUILD_SUFFIX: cu124
-  VLLM_PR: 16104
   VLLM_REF: v0.8.3
   VLLM_BUILD_VERSION: 0.8.3
 
@@ -72,7 +71,6 @@ jobs:
             XFORMERS_BUILD_VERSION=${{ env.XFORMERS_BUILD_VERSION }}
             FLASHINFER_REF=${{ env.FLASHINFER_REF }}
             FLASHINFER_BUILD_SUFFIX=${{ env.FLASHINFER_BUILD_SUFFIX }}
-            VLLM_PR=${{ env.VLLM_PR }}
             VLLM_REF=${{ env.VLLM_REF }}
             VLLM_BUILD_VERSION=${{ env.VLLM_BUILD_VERSION }}
           cache-from: type=registry,ref=${{ env.GHCR_IMAGE }}:cache-cu${{ env.CUDA_TAG }}-${{ matrix.image_distro }}-${{ matrix.arch }}

Dockerfile

@@ -167,19 +167,13 @@ FROM build-base AS build-vllm
 COPY --from=build-torch /wheels/*.whl wheels/
 RUN uv pip install wheels/*
 
-ARG VLLM_PR
 ARG VLLM_REF=v0.8.3
 ARG VLLM_BUILD_VERSION=0.8.3
 ENV BUILD_VERSION=${VLLM_BUILD_VERSION:-${VLLM_REF#v}}
 ENV SETUPTOOLS_SCM_PRETEND_VERSION=${BUILD_VERSION:-:}
 RUN git clone https://github.com/vllm-project/vllm.git
 RUN cd vllm && \
-    if [ -n ${VLLM_PR} ]; then \
-        git fetch origin pull/${VLLM_PR}/head:${VLLM_PR} && \
-        git checkout ${VLLM_PR}; \
-    else \
-        git checkout ${VLLM_REF}; \
-    fi && \
+    git checkout ${VLLM_REF} && \
     python use_existing_torch.py && \
     uv pip install -r requirements/build.txt && \
     uv build --wheel --no-build-isolation -o /wheels