data-preparation
Prepare nnU-Net v2-compatible datasets, path variables, dataset.json metadata, conversions, inference input naming, labels/regions/ignore-label conventions, and validation troubleshooting.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- data-preparation
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Prepare nnU-Net v2-compatible datasets, path variables, dataset.json metadata, conversions, inference input naming, labels/regions/ignore-label conventions, and validation troubleshooting.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/conversion-and-validation.md4.5 KB
- 参考文档references/data-formats.md5.5 KB
- 参考文档references/path-setup.md2.5 KB
- 参考文档references/troubleshooting.md5.2 KB
- 脚本scripts/validate_dataset_json.py13.8 KB
03 SKILL.md 正文
data-preparation
Use this sub-skill when the task is to make data acceptable to nnU-Net v2 before planning/preprocessing.
Route here for
- Setting
nnUNet_raw,nnUNet_preprocessed,nnUNet_results, or optionalnnUNet_extTrainer. - Creating
nnUNet_raw/DatasetXXX_Namefolders withimagesTr,labelsTr, optionalimagesTs, anddataset.json. - Naming training or inference inputs with
_0000,_0001, ... channel suffixes and a consistent file ending. - Authoring or checking
dataset.json, including labels, regions,regions_class_order, ignore labels, and reader/writer overrides. - Converting Medical Segmentation Decathlon or old nnU-Net v1 datasets into nnU-Net v2 format.
- Diagnosing validation errors before
nnUNetv2_plan_and_preprocess.
Do not handle here
- Planning fingerprints, experiment planners, or preprocessing outputs: use
planning-preprocessing. - Trainer classes, folds, losses, schedules, or training commands: use
training-configuration. - Prediction, ensembling, evaluation, or postprocessing after a trained model exists: use
inference-evaluation.
Start with these references
- Dataset layout, naming,
dataset.json, regions, ignore labels, and inference input format:references/data-formats.md. - Storage path environment variables and lazy path failure behavior:
references/path-setup.md. - MSD/v1 conversion command patterns and preflight validation:
references/conversion-and-validation.md. - Common failure modes and fixes:
references/troubleshooting.md.
Bundled helper
Run the lightweight validator before planning/preprocessing:
python sub-skills/data-preparation/scripts/validate_dataset_json.py /path/to/nnUNet_raw/Dataset123_MyDataset
Add --check-files to verify that imagesTr and labelsTr filenames match dataset.json channel count and file ending. The helper checks metadata and names only; it does not inspect image geometry or label voxel values.
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).