Skip to content

cp: Update NDD (nemo data-designer) version (1755) into r1.2.0#1849

Merged
ayushdg merged 1 commit intor1.2.0from
cherry-pick-1755-r1.2.0
Apr 22, 2026
Merged

cp: Update NDD (nemo data-designer) version (1755) into r1.2.0#1849
ayushdg merged 1 commit intor1.2.0from
cherry-pick-1755-r1.2.0

Conversation

@svcnvidia-nemo-ci
Copy link
Copy Markdown
Contributor

beep boop [🤖]: Hi @huvunvidia 👋,

we've cherry picked #1755 into  for you! 🚀

Please review and approve this cherry pick by your convenience!

* update ndd version

Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>

* Update pyproject.toml

Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Signed-off-by: Huy Vu <86480512+huvunvidia@users.noreply.github.com>
Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>

* Update fsspec override comment to clarify nemo-toolkit[asr] conflict

nemo-toolkit[asr]==2.7.2 pins fsspec==2024.12.0, which conflicts with
data-designer-engine==0.5.5 requiring fsspec>=2025.3.0.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>

* Regenerate uv.lock with updated fsspec override

Updates lock file to reflect fsspec>=2025.3.0 override needed due to
nemo-toolkit[asr]==2.7.2 pinning fsspec==2024.12.0.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>

* Cap protobuf <7.0 to fix Ray Serve FieldDescriptor API breakage

protobuf 7.x dropped the .label attribute from the C extension
FieldDescriptor, breaking ray/serve/_private/config.py:75. Cap to <7.0
in both constraint-dependencies and override-dependencies so uv resolves
to 6.x (was bumping to 7.34.1 after lock regeneration).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>

* Cap vllm <0.16.0 to fix Ray Serve LLMServer deployment failure

ray[llm]==2.54 is incompatible with vllm>=0.16.0 which renamed internal
modules (translations -> speech_to_text), causing LLMServer to hit
DEPLOY_FAILED. Cap vllm to <0.16.0 until ray[llm] adds support.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>

* Address comments

Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>

* Remove hard pins and rely on base uvlock deps

Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>

* Forgot to update ray-serve pin

Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>

---------

Signed-off-by: Ayush Dattagupta <ayushdg95@gmail.com>
Signed-off-by: Huy Vu <86480512+huvunvidia@users.noreply.github.com>
Co-authored-by: Huy Vu2 <huvu@login-eos02.eos.clusters.nvidia.com>
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Ayush Dattagupta <ayushdg95@gmail.com>
Signed-off-by: NeMo Bot <nemo-bot@nvidia.com>
@copy-pr-bot
Copy link
Copy Markdown

copy-pr-bot Bot commented Apr 22, 2026

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@svcnvidia-nemo-ci
Copy link
Copy Markdown
Contributor Author

/ok to test a153a53

@greptile-apps
Copy link
Copy Markdown
Contributor

greptile-apps Bot commented Apr 22, 2026

Greptile Summary

This cherry-pick (from #1755) upgrades data-designer from 0.4.0 to 0.5.5 on the r1.2.0 branch, along with the necessary dependency guard-rails: a vllm<0.16.0 upper-bound for inference_server, a protobuf<7.0 cap to avoid a ray-serve FieldDescriptor API breakage, and an fsspec>=2025.3.0 override required by the new data-designer-engine. The lock file is fully regenerated and consistent with the declared changes.

Confidence Score: 5/5

Safe to merge — straightforward dependency version bump with well-motivated upper bounds and a consistent lock file regeneration.

All changes are dependency metadata (pyproject.toml + uv.lock). Every new constraint is accompanied by an explanatory comment, the sdg_cuda12 extra correctly pulls in sdg_cpu (so the data-designer pin propagates), and no application logic is touched. No P0/P1 findings.

No files require special attention.

Important Files Changed

Filename Overview
pyproject.toml Bumps data-designer from 0.4.0 to 0.5.5; adds vllm<0.16.0 cap for inference_server; tightens protobuf to <7.0; adds fsspec>=2025.3.0 override — all changes are well-commented and internally consistent
uv.lock Lock file regenerated to reflect data-designer 0.5.5 (and config/engine sub-packages), duckdb bump 1.4.4→1.5.2 as transitive dep, and updated fsspec/protobuf constraints — consistent with pyproject.toml changes

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[pyproject.toml changes] --> B[data-designer==0.5.5]
    A --> C[vllm<0.16.0\nx86_64 non-Darwin only]
    A --> D[protobuf>=5.29.6,<7.0\nCVE fix + ray serve guard]
    A --> E[fsspec>=2025.3.0 override]

    B --> F[data-designer-config 0.5.5]
    B --> G[data-designer-engine 0.5.5]
    G --> H[+ chardet\n+ fsspec\n- litellm]
    F --> I[+ pillow]

    G --> J[duckdb 1.5.2\nTransitive bump]

    E --> K[Overrides nemo-toolkit\nfsspec==2024.12.0 pin]
Loading

Reviews (1): Last reviewed commit: "Update NDD (nemo data-designer) version ..." | Re-trigger Greptile

@ayushdg ayushdg merged commit 52d57ce into r1.2.0 Apr 22, 2026
50 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants