test-maintain-template
Maintain Lightning-Hydra-Template tests, CI, Makefile targets, package metadata, smoke profiles, RunIf skips, and package renames.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- test-maintain-template
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Maintain Lightning-Hydra-Template tests, CI, Makefile targets, package metadata, smoke profiles, RunIf skips, and package renames.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/package-renaming.md2.1 KB
- 参考文档references/testing-ci-reference.md2.9 KB
- 参考文档references/troubleshooting.md2.0 KB
- 脚本scripts/select_smoke_tests.py2.2 KB
03 SKILL.md 正文
Test and Maintain the Template
Use this sub-skill when the task is about repository maintenance rather than model training itself.
Triggers
Read this sub-skill for tasks about:
- Choosing pytest commands, no-network smoke tests, slow tests, GPU tests, or sweep tests.
tests/conftest.py,tests/test_configs.py,tests/test_train.py,tests/test_eval.py,tests/test_sweeps.py, ortests/helpers/RunIf.make test,make test-full,make format,pre-commit, and GitHub Actions.- Updating
setup.py, console scripts, imports, coverage paths, and config targets after renaming the defaultsrcpackage. - CI failures from missing optional dependencies, Hydra global state, strict markers, or data downloads.
Quick workflow
- Classify tests before running them:
python <this-skill>/sub-skills/test-maintain-template/scripts/select_smoke_tests.py --profile offline python <this-skill>/sub-skills/test-maintain-template/scripts/select_smoke_tests.py --profile full - For a required no-training smoke, run:
pytest tests/test_configs.py -q - Do not assume
make testis no-network. It expands topytest -k "not slow", and the non-slow fast-dev training test can still download MNIST. - When renaming packages or moving entry files, read package renaming and run the target checker from customize-data-model.
- For CI, keep optional GPU/sweep/logger tests guarded by
RunIfor separate profiles.
Read references
- Testing and CI reference: pytest fixtures, native test classification, Makefile targets, CI matrix, and safe command profiles.
- Package renaming: update checklist for
srcpackage renames and setup/coverage/config targets. - Troubleshooting: Hydra state, MNIST downloads, optional skips, strict markers,
sh, logger credentials, and CI surprises.
Bundled script
- scripts/select_smoke_tests.py: prints recommended pytest commands and risk notes for offline, quick, full, gpu, sweep, and rename profiles.
Boundaries
- For Hydra config content and sweeps as experiment design, use configure-experiments.
- For train/eval semantics and checkpoints, use train-evaluate.
- For data/model implementation and
_target_import checks, use customize-data-model.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).