software-engineering/spark-vllm-docker
2
0
Fork 0
Code Issues Actions Packages Projects Releases Wiki Activity

Updated MXFP4 build to use fresh repo references

Browse Source
This commit is contained in:
Eugene Rakhmatulin
2026-02-18 13:35:09 -08:00
parent 5b2313dddb
commit bd3f45f920
2 changed files with 3 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
Dockerfile.mxfp4
View File

@@ -98,7 +98,7 @@ ARG FLASHINFER_REPO=https://github.com/christopherowen/flashinfer.git
ARG CUTLASS_REPO=https://github.com/christopherowen/cutlass.git
ARG FLASHINFER_SHA=f349e52496a72a00d8c4ac02c7a1e38523ff7194
ARG CUTLASS_SHA=11af7f02ab52c9130e422eeb4b44042fbd60c083
ARG CUTLASS_SHA=7554e4a32b15929289aa80e8bc8a4e5f6f779407
RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
uv pip install nvidia-nvshmem-cu13 "apache-tvm-ffi<0.2"
@@ -167,7 +167,7 @@ ARG CACHEBUST_VLLM=1
ARG VLLM_REPO=https://github.com/christopherowen/vllm.git
# Git reference (branch, tag, or SHA) to checkout
ARG VLLM_SHA=459541683f2d8c21f9c0e2f44954b04f59611cbe
ARG VLLM_SHA=045293d82b832229560ac4a13152a095af603b6e
# 4. Smart Git Clone (Fetch changes instead of full re-clone)
# We mount a cache at /repo-cache. This directory persists on your host machine.