Skip to content

Add MaxText and Raiden optional dependencies to Dockerfile #2013

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
SurbhiJainUSC wants to merge 1 commit into
base: main
Choose a base branch
from
+37 −2
Open

Add MaxText and Raiden optional dependencies to Dockerfile #2013

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
22 changes: 22 additions & 0 deletions Dockerfile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,7 @@ ENV PATH="/opt/venv/bin:$PATH"
# Upgrade pip
RUN pip install --upgrade pip

RUN pip install uv
RUN pip install git+https://github.com/ayaka14732/jax-smi.git
# If you encounter a checkpoint issue, try using following old version of pathways-utils.
# RUN pip install git+https://github.com/AI-Hypercomputer/pathways-utils.git@b72729bb152b7b3426299405950b3af300d765a9#egg=pathwaysutils
Expand All @@ -37,6 +38,27 @@ RUN pip install -e .

RUN bash /app/scripts/install_tunix_vllm_requirement.sh

# Build argument to conditionally install MaxText dependencies
ARG INSTALL_MAXTEXT=false

# Install MaxText specific dependencies conditionally
RUN if [ "$INSTALL_MAXTEXT" = "true" ]; then \
uv pip install -r /app/requirements/maxtext_requirements.txt --torch-backend=cpu; \
fi

# Build argument to conditionally install Raiden weight sync dependencies
ARG INSTALL_RAIDEN=false

# Install Raiden specific dependencies conditionally
RUN if [ "$INSTALL_RAIDEN" = "true" ]; then \
if [ -d "/app/raiden_wheels" ] && ls /app/raiden_wheels/*.whl 1>/dev/null 2>&1; then \
pip install --force-reinstall --no-deps /app/raiden_wheels/*.whl; \
else \
pip install keyrings.google-artifactregistry-auth && \
pip install tpu-raiden-jax --extra-index-url https://us-python.pkg.dev/cloud-tpu-inference-test/tpu-raiden/simple/; \
Comment thread
SurbhiJainUSC marked this conversation as resolved.
fi; \
fi

# Build argument to conditionally install DeepSWE evaluation dependencies
ARG INSTALL_DEEPSWE_DEPS=false

Expand Down
14 changes: 14 additions & 0 deletions requirements/maxtext_requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,14 @@
maxtext @ git+https://github.com/AI-Hypercomputer/maxtext.git@8c2e29218c01b6287ba85e8d0dc9555561f7634c
maxtext-vllm-adapter @ git+https://github.com/AI-Hypercomputer/maxtext.git@8c2e29218c01b6287ba85e8d0dc9555561f7634c#subdirectory=src/maxtext/integration/vllm
aqtp
tokamax>=0.0.4
drjax>=0.1.4
ml-goodput-measurement
cloud-accelerator-diagnostics
cloud-tpu-diagnostics
google-cloud-mldiagnostics>=0.5.10
google-cloud-monitoring
einshape
typeguard
xprof
cheroot
3 changes: 1 addition & 2 deletions requirements/special_requirements.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,5 +2,4 @@
# --find-links https://storage.googleapis.com/jax-releases/libtpu_releases.html
# --pre

tpu-inference @ git+https://github.com/vllm-project/tpu-inference.git@12eed6e661a1ffacbce9d107a9894c1a471e7479

@jiangyangmu jiangyangmu Aug 28, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Bump this version should be careful, we need to run package test and nightly regression test before bump.

Do you depend on changes in the updated version ?

@SurbhiJainUSC SurbhiJainUSC Aug 28, 2026

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, we need these weight sync changes in tpu-inference: vllm-project/tpu-inference@4a89516

@wang2yn84 wang2yn84 Aug 28, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

When you update tpu-inference, please also update vllm, and the vllm nightly image tag in https://github.com/google/tunix/tree/main/.github/workflows. So the correct way of doing this is identify which vllm docker image you'd like to use, then apply the same commit to these requirement files.


tpu-inference @ git+https://github.com/vllm-project/tpu-inference.git@4a8951606026d5c0307554bc4ed93f4690c4640a
Loading