data-preparation
Prepare, validate, register, and troubleshoot Dexbotic DexData datasets and safe dataset conversions.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- data-preparation
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Prepare, validate, register, and troubleshoot Dexbotic DexData datasets and safe dataset conversions.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/conversion-boundaries.md2.1 KB
- 参考文档references/dexdata-and-registration.md3.6 KB
- 参考文档references/troubleshooting.md2.1 KB
- 脚本scripts/convert_lerobot_to_dexdata.py3.2 KB
- 脚本scripts/validate_dexdata.py3.6 KB
03 SKILL.md 正文
Dexbotic data preparation
Use this route when the task is about converting demonstrations, designing a DexData JSONL schema, registering a dataset, computing action/state normalization metadata, or diagnosing data-loader failures. Keep model/trainer selection in training, HTTP policy serving in inference-serving, and physical robot topology in evaluation-deployment.
Operating sequence
- Decide the action/state convention before conversion: dimensionality, delta versus absolute dimensions, periodic dimensions, gripper convention, and camera order.
- Produce one JSON object per frame in an episode JSONL file. Keep image/video references and frame indices aligned with the same frame.
- Validate paths, finite numeric values, frame indices, prompt presence, and consistent state/action dimensions with
scripts/validate_dexdata.pybefore registering the dataset. - Register a dataset dictionary with
register_dataset(...), includingannotations, optionaldata_path_prefix, positivefrequency, and action metadata. Import the registration module before constructingDexDataset. - Generate or load
norm_stats.jsonalongside the checkpoint/data contract; never silently reuse statistics from another action space. - Run a small loader inspection before expensive training. A missing image, malformed JSONL line, or bad action horizon is a data issue, not a model issue.
Read DexData and registration, conversion boundaries, and troubleshooting for details. The bundled validator is intentionally local, deterministic, and read-only unless an explicit report path is supplied.
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).