tasks-datasets-and-envs
Use Habitat-Lab task, dataset, environment, Gym, and vectorized environment workflows safely.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- tasks-datasets-and-envs
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Use Habitat-Lab task, dataset, environment, Gym, and vectorized environment workflows safely.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/api-reference.md6.8 KB
- 参考文档references/data-formats.md6.6 KB
- 参考文档references/troubleshooting.md6.7 KB
- 参考文档references/workflows.md7.9 KB
- 脚本scripts/habitat_config_smoke.py8.6 KB
03 SKILL.md 正文
Habitat-Lab Tasks, Datasets, and Environments
Use this sub-skill when the user needs to load Habitat task configs, inspect dataset and episode layout, construct habitat.Env or habitat.RLEnv, wrap tasks as Gym environments, use VectorEnv/ThreadedVectorEnv, or diagnose environment/data/config smoke-check failures.
Read First
- API reference: verified constructor signatures, object relationships, lifecycle calls, Gym conversion behavior, and vectorized-env methods.
- Workflows: config-to-env recipes, Gym wrappers, VectorEnv debug mode, navigation/rearrangement/VLN/EQA patterns, validation steps, and skip conditions.
- Data formats: scene/task dataset layouts,
Episodefields, task-specific episode extensions, DATASETS-derived paths, and tiny-fixture guidance. - Troubleshooting: missing assets, Habitat-Sim/graphics failures, VectorEnv hidden errors, action/observation-space mismatches, Gym wrapper issues, and backend failures.
Bundled Script
Use scripts/habitat_config_smoke.py for safe config inspection. It loads a Habitat config, applies Hydra overrides, prints environment/task/simulator/dataset summaries, and only attempts dataset construction when --make-dataset is explicitly provided. It never downloads data and never creates a simulator by default.
Boundary Reminders
- For install, editable package setup, Hydra search path basics, Habitat-Sim installation, CUDA/graphics setup, and dependency repair, route to
setup-and-configuration. - For RL training, evaluator loops, checkpoints, policies, and Habitat-Baselines trainer configuration, route to
baselines-training-and-evaluation. - For real-time HITL apps, interactive keyboard/gamepad/VR control, and app services, route to
hitl-apps-and-interaction. - For registering custom sensors, actions, tasks, measures, datasets, or simulators, route to
extension-patterns.
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).