cp: fix: deepscaler-24k test reduce to 10 steps to safely finish in 4 hr (1280) into r0.4.0

[chtruong814]

beep boop [🤖]: Hi @terrykong 👋,

we've cherry picked #1280 into  for you! 🚀

Please review and approve this cherry pick by your convenience!

Summary by CodeRabbit

• Tests
  • Reduced steps per run and maximum step thresholds in the LLM test suite.
  • Recalculated run counts and adjusted control flow and gating/check logic to match new limits.
  • Aligned post-run validations and conversions with revised thresholds.
  • No changes to application features or behavior for end-users.

[coderabbitai]

📝 Walkthrough

Walkthrough

Updated test configuration constants in a single shell script: STEPS_PER_RUN and MAX_STEPS were changed from 15 to 10, impacting NUM_RUNS computation and any checks relying on the maximum step threshold in the test flow.

Changes

Cohort / File(s)	Summary
LLM test suite config adjustments tests/test_suites/llm/grpo-deepscaler-1.5b-24K.sh	Updated constants: STEPS_PER_RUN=10 (was 15), MAX_STEPS=10 (was 15); implicitly adjusts NUM_RUNS and related step-gated checks/conversions.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~2 minutes

Possibly related PRs

• fix: deepscaler-24k test reduce to 10 steps to safely finish in 4 hr #1280 — Modifies the same test script to change STEPS_PER_RUN and MAX_STEPS from 15 to 10.

Suggested labels

r0.4.0

Suggested reviewers

• terrykong

Pre-merge checks and finishing touches

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title Check	✅ Passed	The provided title clearly highlights the key change of reducing the deepscaler-24k test to 10 steps, which aligns with the modifications in the changeset, even though it includes extra cherry-pick and branch details that do not obscure the main fix.
Docstring Coverage	✅ Passed	No functions found in the changes. Docstring coverage check skipped.
Test Results For Major Changes	✅ Passed	The PR adjusts two shell-script constants in a test configuration, reducing the number of steps per run from 15 to 10, which is a minor tweak intended to shorten execution time and does not constitute a major feature, breaking change, or numerical/performance-sensitive modification; therefore, the requirement for explicit test results in the description does not apply.

✨ 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 cherry-pick-1280-r0.4.0

---

📜 Recent review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between a334b30 and 6397b3c.

📒 Files selected for processing (1)

• tests/test_suites/llm/grpo-deepscaler-1.5b-24K.sh (1 hunks)

🧰 Additional context used

📓 Path-based instructions (3)

**/*.sh

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

**/*.sh: Follow the Google Shell Style Guide for all shell scripts
Use uv run to execute Python scripts in shell/driver scripts instead of activating virtualenvs and calling python directly
Add the NVIDIA copyright header (with current year) at the top of all shell scripts, excluding tests/ and test-only scripts

Files:

• tests/test_suites/llm/grpo-deepscaler-1.5b-24K.sh

tests/test_suites/llm/*.sh

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

LLM driver script filenames must mirror the YAML base name and follow the same pattern with .sh extension

Files:

• tests/test_suites/llm/grpo-deepscaler-1.5b-24K.sh

tests/test_suites/**

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

Place driver shell scripts and common.env under tests/test_suites// and list nightly tests in tests/test_suites/nightly.txt

Files:

• tests/test_suites/llm/grpo-deepscaler-1.5b-24K.sh

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (5)

• GitHub Check: CI quality check
• GitHub Check: Lint check
• GitHub Check: Lint check
• GitHub Check: Post submodule check comment / Comment on PR
• GitHub Check: Post automodel integration comment / Comment on PR

🔇 Additional comments (1)

tests/test_suites/llm/grpo-deepscaler-1.5b-24K.sh (1)

7-8: Reduction to 10 steps looks good.

Lowering both STEPS_PER_RUN and MAX_STEPS to 10 keeps the run/count math consistent (NUM_RUNS still resolves to 1) while shortening the total training window. Downstream checkpoints and metric assertions remain aligned via $MAX_STEPS.

---

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.}