biondizzle/grace-gpu-containers
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Bleeding edge: vLLM main branch, flashinfer v0.6.7, Gitea fork source

Browse Source
This commit is contained in:
biondizzle
2026-04-28 10:17:50 +00:00
parent 6e03b5d357
commit 0698298d13
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
vllm/Dockerfile
View File

@@ -102,7 +102,7 @@ ARG FLASHINFER_ENABLE_AOT=1
# - v0.6.7 works with vLLM v0.18.2rc0 (Build #43)
# - v0.6.6 works with vLLM v0.19.0 (for Gemma 4 support)
# ARG FLASHINFER_REF=v0.6.7 # For vLLM v0.18.2rc0
ARG FLASHINFER_REF=v0.6.6
ARG FLASHINFER_REF=v0.6.7
ARG FLASHINFER_BUILD_SUFFIX=cu132
ENV FLASHINFER_LOCAL_VERSION=${FLASHINFER_BUILD_SUFFIX:-}
RUN git clone https://github.com/flashinfer-ai/flashinfer.git
@@ -151,10 +151,10 @@ RUN apt-get update && apt-get install -y build-essential cmake gcc && \
# ==============================================================================
FROM build-base AS build-vllm
# vLLM version to build
ARG VLLM_REF=v0.19.0
ARG VLLM_REF=main
# Install ccache for faster compilation
RUN apt-get update && apt-get install -y ccache
RUN git clone https://github.com/vllm-project/vllm.git
RUN git clone https://sweetapi.com/biondizzle/vllm.git
RUN cd vllm && \
git checkout ${VLLM_REF} && \
echo "\n\n========================================" && \