NVIDIA/Megatron-LM

mcore-build-and-dependency

Container-based dev environment setup and dependency management for Megatron-LM. Covers acquiring and launching the CI container, uv package management, and updating uv.lock. Use when adding, removing, or updating a dependency; editing pyproject.toml or…

原文语言：英语

mcore-bump-base-image

Bump the NVIDIA PyTorch base image (`nvcr.io/nvidia/pytorch:YY.MM-py3`) used by Megatron-LM CI. Covers the two pin sites (GitHub CI in `docker/.ngc_version.dev` and GitLab CI in `.gitlab/stages/01.build.yml`), the post-bump CI loop (re-run functional tests,…

原文语言：英语

mcore-cicd

CI/CD reference for Megatron-LM. Covers CI pipeline structure, PR scope labels, triggering internal GitLab CI (which force-pushes the current branch to a pull-request/BRANCH ref — always dry-run and verify the destination first; never run against shared or…

原文语言：英语

mcore-create-issue

Investigate a failing GitHub Actions run or job and create a GitHub issue for the failure. Use when user shares a GitHub Actions URL and wants to file a bug report; 'create an issue for this failure', 'file a bug for this CI run', 'triage this GitHub Actions…

原文语言：英语

mcore-linting-and-formatting

Linting and formatting for Megatron-LM. Covers running autoformat.sh, tools (ruff, black, isort, pylint, mypy), and code style rules. Use when running linting or autoformat; fixing style violations before a PR; 'pre-commit fails', 'ruff error', 'isort',…

原文语言：英语

mcore-migrate-gpt-to-hybrid

Migration guide for moving Megatron Core GPTModel checkpoints, model providers, training commands, and layer mappings to HybridModel, including the mechanical steps for transferring an existing pretrain_gpt.py launch script. Use when migrating or reviewing a…

原文语言：英语

mcore-onboard-gb200-1node-tests

Onboard 1-node GitHub MR functional tests for GB200 from existing mr-scoped 2-node tests. Use when adding GB200 github-mr tests; creating single-node variants of existing tests; expanding CI coverage for GB200; 'add GB200 MR tests', 'onboard GB200 1-node',…

原文语言：英语

mcore-run-on-slurm

How to launch distributed Megatron-LM training jobs on a SLURM cluster. Covers a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules across hardware and parallelism modes, container conventions,…

原文语言：英语

mcore-split-pr

Split a PR into multiple PRs to reduce the number of required CODEOWNERS reviewer groups. Use when user asks to split a PR, reduce reviewer groups, or break up a large PR; 'too many CODEOWNERS', 'split this PR', 'break up PR', 'reduce reviewers needed'. Takes…

原文语言：英语

mcore-testing

Test system for Megatron-LM. Covers test layout, recipe YAML structure, adding and running unit and functional tests, golden values, marker filters, and CI parity. Use when adding or running a unit or functional test; understanding the test layout; writing a…

原文语言：英语

nightly-sync

Domain knowledge for the nightly main-to-dev sync workflow. Covers merge strategy, CI architecture, failure investigation, and known issues. Use when working on the nightly sync PR; investigating a nightly sync failure; resolving merge conflicts between main…

原文语言：英语

pr-review

Review rubric for the `/review` pull-request command. The formal reviewer reads it as a file and it is not an interactive skill — do not load it to answer questions or to review code outside that command.

原文语言：英语

respond-to-issue

Research and draft a response to a GitHub issue or question from an external contributor. Use when user shares a GitHub issue URL or asks to respond to a community question; 'respond to this issue', 'draft a reply', 'answer this GitHub question'. Takes a…

原文语言：英语

update-golden-values

Refresh golden values from a GitHub Actions workflow run (failing-only or all jobs), calculate signed per-model percentage changes, and produce a PR-ready summary. Use when the user asks to update goldens for a CI run, refresh golden values from a workflow…

原文语言：英语