vision-multimodal
Operate AutoTrain Advanced image classification, image regression, object detection, and VLM app/API/config workflows.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- vision-multimodal
- license
- Apache 2.0
- metadata
- { "disco-role": "operating", "parent-skill": "autotrain-advanced" }
- description
- Operate AutoTrain Advanced image classification, image regression, object detection, and VLM app/API/config workflows.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/data-formats.md2.0 KB
- 参考文档references/troubleshooting.md2.1 KB
- 参考文档references/workflows.md2.5 KB
- 脚本scripts/validate_vision_data.py9.0 KB
03 SKILL.md 正文
AutoTrain vision and multimodal workflows
Use this sub-skill for image classification, image regression/scoring, object detection, and VLM dataset/task flows.
Supported entry points
autotrain image-classification --helpautotrain image-regression --helpautotrain object-detection --help- YAML aliases:
image-classification,image-regression,image-scoring,object-detection,image-object-detection,vlm:captioning,vlm:vqa - App/API task keys:
image-classification,image-regression,image-object-detection,vlm:captioning,vlm:vqa
Important: there is no top-level autotrain vlm command in this checkout. Route VLM through the app/API/config flow.
Data-layout summary
| Task | Local layout |
|---|---|
| image classification | directory with at least two class subfolders; each class folder has at least five jpg/jpeg/png files and no extra files/subfolders |
| image regression | directory with at least five images plus metadata.jsonl containing file_name and target |
| object detection | directory with at least five images plus metadata.jsonl containing file_name and objects |
| VLM | directory with at least five images plus metadata.jsonl containing file_name and all mapped text/prompt columns |
Use scripts/validate_vision_data.py for bounded local layout checks.
Safe validation sequence
- Inspect relevant CLI help with the root
inspect_cli.pyhelper. - Validate YAML with the root
validate_config.pyhelper. - Validate local image folders/metadata with
scripts/validate_vision_data.py. - If VLM or UI/API parameters are involved, use
../app-backends/to inspect app/API task params and backend auth. - Launch only after data layout, model id, task alias, backend, and Hub credentials are explicit.
References
references/workflows.md— route patterns and launch/check examples.references/data-formats.md— local folder and metadata schemas.references/troubleshooting.md— image file counts, metadata parsing, VLM route, and backend issues.
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).