auto-ml
Use Igel's AutoKeras-backed Auto-ML helpers for image, text, and structured-data task selection, IgelCNN train/evaluate/predict behavior, and current Auto-ML CLI caveats.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- auto-ml
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Use Igel's AutoKeras-backed Auto-ML helpers for image, text, and structured-data task selection, IgelCNN train/evaluate/predict behavior, and current Auto-ML CLI caveats.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/api-reference.md6.1 KB
- 参考文档references/troubleshooting.md5.5 KB
- 参考文档references/workflows.md5.1 KB
- 脚本scripts/inspect_auto_ml.py6.8 KB
03 SKILL.md 正文
Auto-ML
Use this sub-skill when the task is about Igel's AutoKeras path: IgelCNN, igel.auto.models.Models.get, image classification/regression, text classification/regression, structured-data AutoKeras task names, or the documented-but-missing igel auto-train command.
Start here:
- Use references/api-reference.md for exact
IgelCNNconstructor/method behavior, save/load assumptions, and supported task names. - Use references/workflows.md for the safe operating sequence, directory layouts, and bounded examples.
- Use references/troubleshooting.md when imports, task names, data layout, artifacts, or CLI expectations fail.
- Use scripts/inspect_auto_ml.py for an import/signature/task-selector smoke check that does not train, download data, or require the original source checkout.
Route boundaries:
- Route classic CSV/tabular
igel fit,evaluate,predict,experiment,models,metrics, and ONNX/export work to tabular-workflows; those are not AutoKeras workflows. - Route repo-wide skill selection back to the root router when the user is not clearly asking for Auto-ML.
- Route FastAPI serving, Docker, GUI, and client-request tasks away from this sub-skill; this sub-skill only covers the AutoKeras-backed training/evaluation/prediction helpers and task registry.
- Treat AutoKeras training as potentially expensive. Prefer import/signature and task-selector checks before starting any run, and never make dataset downloads or long tuning loops the default helper path.
- Do not rely on
igel auto-trainas a real current Click CLI command. The docs mention it, but the current CLI surface does not expose that command; use the programmaticIgelCNNpath or route to classic tabular commands when appropriate.
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).