Skip to content
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
28 changes: 28 additions & 0 deletions docker/sglang/v0.5.12/Dockerfile
Original file line number Diff line number Diff line change
@@ -0,0 +1,28 @@
ARG SGLANG_IMAGE=598726163780.dkr.ecr.us-west-2.amazonaws.com/tgl:v0.5.12-cu130-43a42c1bd9-dirty-warm-20260526
FROM ${SGLANG_IMAGE} AS sglang

WORKDIR /root/

COPY patches/sglang/v0.5.12/sglang.patch /sgl-workspace/
RUN cd /sgl-workspace/sglang && \
git apply /sgl-workspace/sglang.patch && \
rm /sgl-workspace/sglang.patch && \
cd python && pip install -e .

COPY . /root/torchspec
RUN cd /root/torchspec && pip install --no-cache-dir -e ".[fa]"

# TorchSpec's dependency pulls the generic CUDA 12 Mooncake wheel. Replace it
# with the CUDA 13 wheel that matches this image family.
RUN pip uninstall -y mooncake-transfer-engine || true && \
pip install --no-cache-dir --no-deps --force-reinstall \
mooncake-transfer-engine-cuda13==0.3.11.post1

RUN chmod 755 /usr/local/lib/python3.12/dist-packages/mooncake/mooncake_master || true
RUN if [ -f /usr/local/lib/python3.12/dist-packages/mooncake/cli.py ]; then \
sed -i 's/os.chmod(bin_path, 0o755)/pass/' /usr/local/lib/python3.12/dist-packages/mooncake/cli.py; \
fi

RUN python3 -c "from sglang.srt.server_args import ServerArgs; from sglang.srt.speculative.spec_training_info import SpecTrainingInfo; assert 'enable_spec_training_mooncake' in ServerArgs.__dataclass_fields__; assert 'enable_aux_hidden_states' in ServerArgs.__dataclass_fields__"

WORKDIR /root/torchspec
Loading
Loading