Separate CI job for Megatron GPU tests

[kevalmorabia97]

What does this PR do?

[Short term]: Megatron based tests take a long time often resulting in CICD timeout. Splitting megatron tests into a dedicated CICD job for faster overall CI/CD run
[Mid/Long term]: Run all megatron gpu tests using torchrun instead of pytest so all dist processes are already created and all individual tests no longer need to setup and destroy their processes which adds a lot of overhead per test

Testing

• 1-GPU CI/CD passing (on this PR)
• 2-GPU CI/CD passing (on nightly run - manually triggered): https://github.com/NVIDIA/Model-Optimizer/actions/runs/22000517688

[coderabbitai]

📝 Walkthrough

Walkthrough

Restructures GPU testing infrastructure by introducing matrix-based parallel execution for multiple test variants, reducing job timeouts to 90 minutes, separating Megatron-specific test configuration into dedicated environments, and consolidating shared test fixtures.

Changes

Cohort / File(s)	Summary
GitHub Workflows .github/workflows/gpu_tests.yml	Implements matrix-based parallel GPU test execution with example variants (py312-cuda12-gpu, py312-cuda12-gpu-megatron), applies fail-fast: false strategy, and reduces job timeouts from 120–150 minutes to 90 minutes.
Megatron Test Configuration tests/gpu_megatron/_extensions, tests/gpu_megatron/torch/conftest.py	Adds path references to shared GPU test fixtures at ../gpu/_extensions/ and ../../gpu/torch/conftest.py to enable fixture reuse across test suites.
Tox Environment Configuration tox.ini	Removes megatron-core from common cuda12-gpu testenv, creates dedicated [testenv:{py310,py311,py312}-cuda12-gpu-megatron] with megatron-core pre-installation, removes Eagle-3 dependencies, and adjusts test execution paths to target tests/gpu_megatron.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~12 minutes

🚥 Pre-merge checks | ✅ 4

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The pull request title accurately captures the main objective: separating Megatron GPU tests into a dedicated CI job, which is reflected in all modified files (.github/workflows/gpu_tests.yml, tox.ini) and the PR description.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Merge Conflict Detection	✅ Passed	✅ No merge conflicts detected when merging into main
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch kmorabia/split-mcore-gpu-tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

.github/workflows/gpu_tests.yml (1)

42-47: ⚠️ Potential issue | 🔴 Critical

Bug: tests/gpu_megatron/** is missing from the changed-files filter.

Changes made only under tests/gpu_megatron/ won't trigger the PR GPU tests because this path isn't included in the file-change detection. The gpu-tests-pr job will be skipped.

Proposed fix

          files: |
            .github/workflows/gpu_tests.yml
            modelopt/**
            tests/gpu/**
+           tests/gpu_megatron/**
            tox.ini
            setup.py

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 73.73%. Comparing base (95511a0) to head (7f8ccaf).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #888   +/-   ##
=======================================
  Coverage   73.73%   73.73%           
=======================================
  Files         199      199           
  Lines       21165    21165           
=======================================
  Hits        15606    15606           
  Misses       5559     5559           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jenchen13]

i think we need to fix the export unit test here to pass in the HF model name for the tokenizer (in export_mcore_gpt_to_hf second argument)