[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78XAXHH87ZCE4N8KX603BYH":3,"skill-same:01M78XAXHH87ZCE4N8KX603BYH":70},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":32,"install":33,"resources":45},"01M78XAXHH87ZCE4N8KX603BYH","datasets-and-configuration",null,"Prepare, select, and validate Gaussian-SLAM RGB-D datasets and YAML configurations for a safe GPU run.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"MIT","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: datasets-and-configuration\ndescription: \"Prepare, select, and validate Gaussian-SLAM RGB-D datasets and\n  YAML configurations for a safe GPU run.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: MIT\n---\n\n# Datasets and configuration\n\nUse this sub-skill when a Gaussian-SLAM run needs an RGB-D scene, a dataset\nalias, a scene-specific YAML file, or a preflight check. It covers the data\ncontract implemented by `src\u002Fentities\u002Fdatasets.py` and the recursive YAML\nmerge behavior in `src\u002Futils\u002Fio_utils.py`. It does **not** own SLAM\noptimization, tracking\u002Fmapping choices, or evaluation metrics.\n\n## Applicability and operating contract\n\n- A run must use one of the exact dataset aliases `replica`, `tum_rgbd`,\n  `scan_net`, or `scannetpp`. These are case-sensitive; `scannet`, `ScanNet`,\n  `tum-rgbd`, and `scannet_pp` are not aliases accepted by `get_dataset`.\n- Start from a dataset default config and use a per-scene config with\n  `inherit_from`. The effective configuration must contain the top-level\n  `project_name`, `dataset_name`, `checkpoint_path`, `use_wandb`, `frame_limit`,\n  `seed`, `mapping`, `tracking`, `cam`, and `data` keys. A per-scene\n  `data.frame_limit` is also accepted and is the value passed to the dataset\n  loader when present (the supplied ScanNet++ scenes use this form).\n- Run the bundled validator before importing CUDA-heavy runtime modules:\n\n  ```bash\n  python skills\u002Fdisco\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration\u002Fscripts\u002Fvalidate_config.py \\\n      configs\u002FReplica\u002Foffice0.yaml\n  ```\n\n  Add `--require-data --path-base \u003Crepository-root>` when the data must be\n  present now rather than merely checking the YAML contract. Multiple explicit\n  config paths may be supplied. The validator uses PyYAML and standard Python\n  modules only; it does not import PyTorch, OpenCV, Open3D, or custom CUDA\n  extensions.\n- `run_slam.py` accepts a config path followed by optional `--input_path` and\n  `--output_path` overrides. Those overrides replace the values under `data`;\n  validate the final values, not only the YAML defaults.\n\n## Safe workflow\n\n1. Identify the dataset and exact scene. Confirm that the external dataset's\n   license, download permissions, credentials, and storage budget permit local\n   use. Do not guess a scene's camera model or silently substitute a dataset.\n2. Inspect the corresponding default and scene YAML. Confirm inheritance,\n   dataset alias, `data.input_path`, `data.output_path`, camera values, depth\n   scale, and split\u002Fframe-limit behavior. Check both the run-level\n   `frame_limit` and any loader-level `data.frame_limit` override.\n3. Acquire or mount data using the reference-only guidance in\n   [data-and-downloads.md](references\u002Fdata-and-downloads.md). Never execute a\n   download merely as part of validation.\n4. Validate the effective config and available data. Resolve relative input and\n   output paths against the intended working directory (the source CLI does\n   not rewrite them). Require paired RGB\u002Fdepth\u002Fpose records and the metadata\n   needed by the selected loader.\n5. Treat warnings as a stop-and-review gate before a GPU run. In particular,\n   investigate relative paths, dropped timestamp associations, changed\n   intrinsics, missing split names, and a frame limit larger than the selected\n   sequence.\n6. Only after the preflight passes, launch the repository command with the\n   final explicit paths. Preserve the effective YAML with the run output for\n   reproducibility.\n\nFor field-level rules and effective merge examples, see\n[configuration.md](references\u002Fconfiguration.md). For on-disk records and image,\ndepth, and pose conventions, see [data-formats.md](references\u002Fdata-formats.md).\nFor download prerequisites and licensing boundaries, see\n[data-and-downloads.md](references\u002Fdata-and-downloads.md). For failure symptoms\nand recovery, see [troubleshooting.md](references\u002Ftroubleshooting.md).\n\n## Expected observations and checks\n\nA successful preflight reports the resolved inheritance chain, effective\n`dataset_name`, camera dimensions\u002Fintrinsics\u002Fdepth scale, selected split (when\napplicable), and either verified data counts or an explicit “data path not\npresent; YAML-only validation” warning. With `--require-data`, absent paths,\nmissing files, malformed records, unsupported aliases, count mismatches, and\nunresolved inheritance are failures (exit status 1).\n\nThe validator is intentionally conservative: it does not modify YAML, create\noutput directories, download data, initialize wandb, allocate a GPU, or import\nthe Gaussian-SLAM package. A passing validator proves configuration and\nlayout preconditions only; it is not evidence that CUDA extensions, tracking,\nmapping, or evaluation will succeed.\n\n## Difficult synthetic usability cases\n\nUse these cases during integrated skill verification; keep fixtures and reports\noutside this runtime skill tree:\n\n1. **Inheritance and path trap:** create `nested\u002Fscene.yaml` inheriting\n   `base.yaml`, run validation from a different current directory, and give the\n   scene a relative `data.input_path`. The validator must load the sibling base\n   by path relative to `scene.yaml`, flag the data path as relative, and reject\n   a second fixture whose base path forms an inheritance cycle.\n2. **Silent frame drop trap:** create a TUM fixture with valid-looking RGB and\n   depth lists but one pose timestamp more than `0.08` seconds from every RGB\n   timestamp; create a ScanNet++ fixture whose selected train split names a\n   frame absent from `transforms_undistorted.json`. Validation must fail with a\n   dataset-specific association\u002Fmetadata message without importing CUDA.\n\n## Linked resources\n\n- [references\u002Fdata-formats.md](references\u002Fdata-formats.md)\n- [references\u002Fconfiguration.md](references\u002Fconfiguration.md)\n- [references\u002Fdata-and-downloads.md](references\u002Fdata-and-downloads.md)\n- [references\u002Ftroubleshooting.md](references\u002Ftroubleshooting.md)\n- [scripts\u002Fvalidate_config.py](scripts\u002Fvalidate_config.py)\n","# Datasets and configuration\n\nUse this sub-skill when a Gaussian-SLAM run needs an RGB-D scene, a dataset\nalias, a scene-specific YAML file, or a preflight check. It covers the data\ncontract implemented by `src\u002Fentities\u002Fdatasets.py` and the recursive YAML\nmerge behavior in `src\u002Futils\u002Fio_utils.py`. It does **not** own SLAM\noptimization, tracking\u002Fmapping choices, or evaluation metrics.\n\n## Applicability and operating contract\n\n- A run must use one of the exact dataset aliases `replica`, `tum_rgbd`,\n  `scan_net`, or `scannetpp`. These are case-sensitive; `scannet`, `ScanNet`,\n  `tum-rgbd`, and `scannet_pp` are not aliases accepted by `get_dataset`.\n- Start from a dataset default config and use a per-scene config with\n  `inherit_from`. The effective configuration must contain the top-level\n  `project_name`, `dataset_name`, `checkpoint_path`, `use_wandb`, `frame_limit`,\n  `seed`, `mapping`, `tracking`, `cam`, and `data` keys. A per-scene\n  `data.frame_limit` is also accepted and is the value passed to the dataset\n  loader when present (the supplied ScanNet++ scenes use this form).\n- Run the bundled validator before importing CUDA-heavy runtime modules:\n\n  ```bash\n  python skills\u002Fdisco\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration\u002Fscripts\u002Fvalidate_config.py \\\n      configs\u002FReplica\u002Foffice0.yaml\n  ```\n\n  Add `--require-data --path-base \u003Crepository-root>` when the data must be\n  present now rather than merely checking the YAML contract. Multiple explicit\n  config paths may be supplied. The validator uses PyYAML and standard Python\n  modules only; it does not import PyTorch, OpenCV, Open3D, or custom CUDA\n  extensions.\n- `run_slam.py` accepts a config path followed by optional `--input_path` and\n  `--output_path` overrides. Those overrides replace the values under `data`;\n  validate the final values, not only the YAML defaults.\n\n## Safe workflow\n\n1. Identify the dataset and exact scene. Confirm that the external dataset's\n   license, download permissions, credentials, and storage budget permit local\n   use. Do not guess a scene's camera model or silently substitute a dataset.\n2. Inspect the corresponding default and scene YAML. Confirm inheritance,\n   dataset alias, `data.input_path`, `data.output_path`, camera values, depth\n   scale, and split\u002Fframe-limit behavior. Check both the run-level\n   `frame_limit` and any loader-level `data.frame_limit` override.\n3. Acquire or mount data using the reference-only guidance in\n   [data-and-downloads.md](references\u002Fdata-and-downloads.md). Never execute a\n   download merely as part of validation.\n4. Validate the effective config and available data. Resolve relative input and\n   output paths against the intended working directory (the source CLI does\n   not rewrite them). Require paired RGB\u002Fdepth\u002Fpose records and the metadata\n   needed by the selected loader.\n5. Treat warnings as a stop-and-review gate before a GPU run. In particular,\n   investigate relative paths, dropped timestamp associations, changed\n   intrinsics, missing split names, and a frame limit larger than the selected\n   sequence.\n6. Only after the preflight passes, launch the repository command with the\n   final explicit paths. Preserve the effective YAML with the run output for\n   reproducibility.\n\nFor field-level rules and effective merge examples, see\n[configuration.md](references\u002Fconfiguration.md). For on-disk records and image,\ndepth, and pose conventions, see [data-formats.md](references\u002Fdata-formats.md).\nFor download prerequisites and licensing boundaries, see\n[data-and-downloads.md](references\u002Fdata-and-downloads.md). For failure symptoms\nand recovery, see [troubleshooting.md](references\u002Ftroubleshooting.md).\n\n## Expected observations and checks\n\nA successful preflight reports the resolved inheritance chain, effective\n`dataset_name`, camera dimensions\u002Fintrinsics\u002Fdepth scale, selected split (when\napplicable), and either verified data counts or an explicit “data path not\npresent; YAML-only validation” warning. With `--require-data`, absent paths,\nmissing files, malformed records, unsupported aliases, count mismatches, and\nunresolved inheritance are failures (exit status 1).\n\nThe validator is intentionally conservative: it does not modify YAML, create\noutput directories, download data, initialize wandb, allocate a GPU, or import\nthe Gaussian-SLAM package. A passing validator proves configuration and\nlayout preconditions only; it is not evidence that CUDA extensions, tracking,\nmapping, or evaluation will succeed.\n\n## Difficult synthetic usability cases\n\nUse these cases during integrated skill verification; keep fixtures and reports\noutside this runtime skill tree:\n\n1. **Inheritance and path trap:** create `nested\u002Fscene.yaml` inheriting\n   `base.yaml`, run validation from a different current directory, and give the\n   scene a relative `data.input_path`. The validator must load the sibling base\n   by path relative to `scene.yaml`, flag the data path as relative, and reject\n   a second fixture whose base path forms an inheritance cycle.\n2. **Silent frame drop trap:** create a TUM fixture with valid-looking RGB and\n   depth lists but one pose timestamp more than `0.08` seconds from every RGB\n   timestamp; create a ScanNet++ fixture whose selected train split names a\n   frame absent from `transforms_undistorted.json`. Validation must fail with a\n   dataset-specific association\u002Fmetadata message without importing CUDA.\n\n## Linked resources\n\n- [references\u002Fdata-formats.md](references\u002Fdata-formats.md)\n- [references\u002Fconfiguration.md](references\u002Fconfiguration.md)\n- [references\u002Fdata-and-downloads.md](references\u002Fdata-and-downloads.md)\n- [references\u002Ftroubleshooting.md](references\u002Ftroubleshooting.md)\n- [scripts\u002Fvalidate_config.py](scripts\u002Fvalidate_config.py)\n",{"name":5,"license":14,"metadata":23,"description":7,"disable-model-invocation":25},{"disco-role":24},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":25},[35,38,41],{"tool":36,"command":37},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill datasets-and-configuration",{"tool":39,"command":40},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill datasets-and-configuration",{"tool":42,"command":43},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill datasets-and-configuration","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration\u002FSKILL.md",[46,52,56,60,64],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fconfiguration.md","REFERENCE","参考文档",5039,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration\u002Freferences\u002Fconfiguration.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Fdata-and-downloads.md",3905,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration\u002Freferences\u002Fdata-and-downloads.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Fdata-formats.md",5663,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration\u002Freferences\u002Fdata-formats.md",{"path":61,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":62,"sourceHtmlUrl":63},"references\u002Ftroubleshooting.md",4155,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration\u002Freferences\u002Ftroubleshooting.md",{"path":65,"resourceType":66,"resourceTypeDesc":67,"sizeBytes":68,"sourceHtmlUrl":69},"scripts\u002Fvalidate_config.py","SCRIPT","脚本",25941,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fgaussian-slam\u002Fsub-skills\u002Fdatasets-and-configuration\u002Fscripts\u002Fvalidate_config.py",[71,79,86,93,100,107],{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M799NYSH1R3RVAHSM2D88Y84","passnet-triton-opt","Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block\u002Fgrid\u002Fwarp 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.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":18},"01M799NS5APEC0N1FZ7XZVW6BN","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.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-solve\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":18},"01M799NNY9BKSM21V5R7QVTBJX","passnet-skill","PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under .\u002Fpass_dir\u002F, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-skill\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":18},"01M799NHH6NP7Z1RK77KGXAN6S","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()\u002Freplacement_args()\u002Freplacement_func() files, when a pass \"failed to match\", or when deciding how to fuse multiple kernels\u002Fops.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-pattern-fusion\u002FSKILL.md",[],[],{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":18},"01M799ND2J4WNC2PGAZ0A06NDF","passnet-orchestrate","Round\u002Fbatch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep\u002Frevert 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.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-orchestrate\u002FSKILL.md",[],[],{"skillNo":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":75,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":113,"publishedAt":18},"01M799N8SBTMY02ZDG2AZF5ZWE","passnet-feedback","Fast, deterministic feedback for PassNet work: pre-flight pattern\u002Fmatch 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).","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-feedback\u002FSKILL.md",[],[]]