artifacts-and-registries
Create, log, consume, download, link, and troubleshoot W&B Artifacts and registry/model-registry flows through Python and CLI.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- artifacts-and-registries
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Create, log, consume, download, link, and troubleshoot W&B Artifacts and registry/model-registry flows through Python and CLI.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/api-reference.md7.8 KB
- 参考文档references/cli-reference.md4.6 KB
- 参考文档references/troubleshooting.md6.6 KB
- 参考文档references/workflows.md5.9 KB
- 脚本scripts/artifact_manifest_smoke.py6.4 KB
03 SKILL.md 正文
W&B Artifacts and Registries
Use this sub-skill when a task involves versioned datasets, models, files, directories, external object-store references, artifact aliases/tags/types, downloads, artifact CLI commands, or linking artifacts into registries/model registries.
Route by task
- Create or log artifacts: Use
references/workflows.md#create-and-log-versioned-artifactsandreferences/api-reference.md#artifact-constructionforwandb.Artifact,add_file,add_dir,add_reference, andRun.log_artifactpatterns. - Consume or download artifacts: Use
references/workflows.md#consume-and-download-artifactsandreferences/api-reference.md#artifact-consumptionforRun.use_artifact,wandb.Api().artifact,Artifact.download,Artifact.file, and path forms. - Use artifact CLI: Use
references/cli-reference.mdforwandb artifact put,get,ls, andcache cleanupsyntax and path expectations. - Link to registries: Use
references/workflows.md#link-artifacts-to-registriesandreferences/api-reference.md#registries-and-linkingforRun.link_artifact,Artifact.link, registry paths, aliases, and public API registry helpers. - Debug failures: Use
references/troubleshooting.mdfor invalid names/types, missing files, credentials, optional storage dependencies, alias/version confusion, finalized mutation errors, and registry path/link errors. - Offline validation helper: Run
python sub-skills/artifacts-and-registries/scripts/artifact_manifest_smoke.py --helpfrom the root skill directory to validate names/types and preview a local manifest without upload.
Boundaries
This sub-skill covers artifact creation, artifact references, artifact lifecycle operations, artifact CLI usage, registry linking, registry lookup, and storage integration gotchas. For general scalar/media metric logging, use the experiment-tracking sub-skill. For large Public API reporting and automation beyond artifacts/registries, use public-api-and-automation. For launch jobs or sweeps that consume artifacts, use sweeps-and-launch after preparing artifacts here.
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).