Skip to content

Restructure trl/ test cases: shared base image and descriptive naming#1000

Merged
mijanur132 merged 2 commits intomainfrom
restructure-trl-test-cases
Feb 27, 2026
Merged

Restructure trl/ test cases: shared base image and descriptive naming#1000
mijanur132 merged 2 commits intomainfrom
restructure-trl-test-cases

Conversation

@KeitaW
Copy link
Collaborator

@KeitaW KeitaW commented Feb 26, 2026

Summary

  • Rename grpo/ to grpo-math-reasoning/ for descriptive test case naming
  • Create a unified trl/Dockerfile with all TRL example dependencies (shared base image)
  • Add a library-level trl/README.md with test case table and build instructions
  • Delete per-test-case grpo.Dockerfile (replaced by shared base)
  • Update all path references in train.sbatch and test case README.md

This establishes the pattern for future TRL test cases to share a single base image rather than duplicating Dockerfiles.

Merge order: This PR should be merged before #997 is updated. Once landed, #997 can rebase onto main and add its new test case (gpt-oss-lora-grpo/) alongside grpo-math-reasoning/, reusing the shared trl/Dockerfile instead of duplicating dependencies.

Test plan

  • Verify git log --follow shows rename history for moved files
  • Verify no references to old grpo/ or grpo.Dockerfile remain
  • Verify trl/Dockerfile builds successfully
  • Verify all srun commands in README and sbatch use trl-base.sqsh and /grpo-math-reasoning/ mount paths

Rename grpo/ to grpo-math-reasoning/ for clarity, create a unified
trl/Dockerfile with all TRL example dependencies, add a HyperPod EKS
Dockerfile extension, and add a library-level README with test case
table. This establishes the pattern for PR #997 and future TRL test
cases to share a single base image rather than duplicating Dockerfiles.
HyperPod EKS support will be added separately when needed.
@KeitaW KeitaW requested a review from mijanur132 February 26, 2026 05:47
@KeitaW KeitaW added the documentation Improvements or additions to documentation label Feb 26, 2026
@paragao paragao self-requested a review February 26, 2026 12:09
Copy link
Collaborator

@mijanur132 mijanur132 left a comment

Choose a reason for hiding this comment

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

Everything looks good.

@mijanur132 mijanur132 merged commit 5f86f92 into main Feb 27, 2026
4 checks passed
@mijanur132 mijanur132 deleted the restructure-trl-test-cases branch February 27, 2026 00:56
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

documentation Improvements or additions to documentation

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants