[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M795ZRJSDSD3YFNCH8N0RCD8":3,"skill-same:01M795ZRJSDSD3YFNCH8N0RCD8":67},{"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":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":6,"ownerHtmlUrl":32,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":33,"install":34,"resources":46},"01M795ZRJSDSD3YFNCH8N0RCD8","beam-and-performance",null,"Scale TensorFlow Datasets generation and reading with Beam, storage, sharding, GCS, and performance controls.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-datasets\u002Fsub-skills\u002Fbeam-and-performance\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: beam-and-performance\ndescription: \"Scale TensorFlow Datasets generation and reading with Beam,\n  storage, sharding, GCS, and performance controls.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: Apache 2.0\n---\n\n# Beam And Performance\n\nUse this sub-skill when the task is to scale TensorFlow Datasets (TFDS) dataset\npreparation or reading with Apache Beam, cloud runners, file formats, sharding,\nGCS data directories, or `tf.data` performance controls.\n\nRoute other work away:\n\n- Generic `tfds.load`, split syntax, metadata inspection, NumPy\u002FDataFrame, and\n  non-Beam loading recipes -> `data-loading`.\n- Full `tfds` CLI syntax outside Beam\u002Ffile-format\u002Fsharding flags ->\n  `cli-workflows`.\n- Creating ordinary `GeneratorBasedBuilder` datasets, `_info`, features,\n  checksums, dummy data, and `DatasetBuilderTestCase` -> `dataset-authoring`.\n- Folder datasets, external TFRecord layouts, HuggingFace\u002FCroissant\u002Fcommunity\n  catalogs, and dataset collections -> `formats-and-community`.\n\n## Evidence-backed API surface\n\nThe TFDS code and environment verification for this skill established these\nruntime facts:\n\n- Beam generation is configured through `tfds.download.DownloadConfig` fields\n  including `beam_runner`, `beam_options`, `max_examples_per_split`,\n  `num_shards`, `min_shard_size`, `max_shard_size`, and\n  `nondeterministic_order`.\n- `tfds build --beam_pipeline_options=...` stores a comma-separated options\n  string; TFDS turns each `key=value` segment into a Beam `PipelineOptions` flag\n  without launching Beam during parsing.\n- Supported prepared-file format names are `tfrecord`, `array_record`,\n  `riegeli`, and `parquet`; TFRecord is the default. Some formats require extra\n  reader\u002Fwriter dependencies or different reading APIs.\n- `tfds.core.BeamBasedBuilder` still exists but is deprecated; current Beam\n  authoring should normally use `GeneratorBasedBuilder` whose\n  `_generate_examples` returns a Beam `PTransform`\u002F`PCollection`.\n- `tfds.beam.ReadFromTFDS` reads an already prepared TFDS dataset inside a Beam\n  pipeline, parallelizing over dataset shards.\n- Dataflow, Flink, private GCS, and service-account credentials are external\n  services; do not run or mutate them without explicit user approval, budget,\n  and credential boundaries.\n\n## Fast workflow chooser\n\n| User intent | Use |\n|---|---|\n| \"Build this huge TFDS dataset locally just to test the builder\" | Direct\u002Flocal Beam runner plus a tiny `--max_examples_per_split` or programmatic `DownloadConfig(max_examples_per_split=...)`; warn that full local builds can exceed RAM\u002Fdisk\u002Ftime. |\n| \"Run TFDS generation on Dataflow\" | Build a GCS-backed `data_dir`, `staging_location`, `temp_location`, `requirements_file`, and `runner=DataflowRunner`; verify project\u002Fbucket\u002Fcredentials before launch. |\n| \"Use Flink\" | Pass `runner=FlinkRunner` plus Flink version\u002Fconfig-dir options; verify Beam\u002FFlink version compatibility externally. |\n| \"Use TFDS as source inside Beam\" | Use `tfds.beam.ReadFromTFDS(builder, split=...)` only after the dataset is prepared; respect subsplit\u002Fbatch limitations. |\n| \"Improve reading throughput\u002FRAM\" | Tune `ReadConfig`, shuffling, cache\u002Fbatch\u002Fprefetch order, decode\u002Fskipped features, file format, and GCS locality. |\n| \"Change shard count\u002Fsize or file format\" | Use generation flags or `DownloadConfig`\u002F`download_and_prepare(file_format=...)`; validate reader compatibility before rebuilding. |\n\n## Standard playbooks\n\n1. **Confirm the generation target.** For multi-config builders, require the\n   exact dataset\u002Fconfig\u002Fversion. Avoid accidental \"all configs\" builds.\n2. **Choose runner and storage.** Direct runner is for tiny local tests. Use a\n   distributed runner for terabyte-scale datasets. Dataflow jobs should usually\n   write TFDS data to a `gs:\u002F\u002F...\u002Ftensorflow_datasets` root and use a GCS temp\n   and staging area.\n3. **Build pipeline options safely.** Use\n   [`scripts\u002Fbeam_options_helper.py`](scripts\u002Fbeam_options_helper.py) to create\n   or validate `--beam_pipeline_options` strings and get warnings without\n   launching Beam.\n4. **Pass options through TFDS.** See\n   [`references\u002Fbeam-workflows.md`](references\u002Fbeam-workflows.md) for CLI and\n   programmatic examples.\n5. **Set deterministic-order policy.** Default Beam writing preserves TFDS's\n   deterministic ordering contract. Use `nondeterministic_order=True` or\n   `--nondeterministic_order` only when faster unordered output is acceptable\n   and downstream code never depends on example order.\n6. **Control storage format and shards.** See\n   [`references\u002Fperformance-and-storage.md`](references\u002Fperformance-and-storage.md)\n   for `--file_format`, `--num_shards`, `--max_shard_size_mb`, and\n   `DownloadConfig` equivalents.\n7. **Tune readers after generation.** Prefer `shuffle_files=True` for large\n   training datasets, tune `ReadConfig` for distributed workers and memory, skip\n   unused or expensive features, and benchmark the final pipeline.\n8. **Triage failures before rerunning expensive jobs.** Use\n   [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) for Beam,\n   GCS, sharding, determinism, and performance symptoms.\n\n## Safe command skeletons\n\nLocal smoke only:\n\n```bash\ntfds build DATASET[\u002FCONFIG] \\\n  --max_examples_per_split=10 \\\n  --data_dir=\u002Ftmp\u002Ftfds-smoke\n```\n\nDataflow option construction without launching a job:\n\n```bash\npython scripts\u002Fbeam_options_helper.py \\\n  --runner DataflowRunner \\\n  --dataset DATASET \\\n  --project GCP_PROJECT \\\n  --gcs-bucket gs:\u002F\u002FBUCKET \\\n  --requirements-file \u002Ftmp\u002Fbeam_requirements.txt \\\n  --print-build-command\n```\n\nProgrammatic generation:\n\n```python\nimport apache_beam as beam\nimport tensorflow_datasets as tfds\n\nflags = [\n    \"--runner=DataflowRunner\",\n    \"--project=GCP_PROJECT\",\n    \"--staging_location=gs:\u002F\u002FBUCKET\u002Fbinaries\",\n    \"--temp_location=gs:\u002F\u002FBUCKET\u002Ftemp\",\n    \"--requirements_file=\u002Ftmp\u002Fbeam_requirements.txt\",\n]\n\ndownload_config = tfds.download.DownloadConfig(\n    beam_options=beam.options.pipeline_options.PipelineOptions(flags=flags),\n    # Only set when unordered output is acceptable:\n    # nondeterministic_order=True,\n)\nbuilder = tfds.builder(\"DATASET\u002FCONFIG\", data_dir=\"gs:\u002F\u002FBUCKET\u002Ftensorflow_datasets\")\nbuilder.download_and_prepare(download_config=download_config)\n```\n\n## Verification expectations\n\nFor this sub-skill, safe verification should avoid full downloads and cloud job\nlaunches unless the user explicitly grants budget and credentials. Prefer:\n\n- `python scripts\u002Fbeam_options_helper.py --help` and representative option\n  construction checks.\n- Import\u002Fsignature checks for `tfds.download.DownloadConfig`, `tfds.ReadConfig`,\n  and `tfds.beam.ReadFromTFDS` in an environment that has the required optional\n  packages.\n- Tiny local Beam\u002Fdummy-data checks only when Apache Beam and TensorFlow are\n  already installed.\n- Documentation of skipped Dataflow\u002FFlink\u002Fprivate-GCS checks as external-service\n  gaps, not failures of local validation.\n","# Beam And Performance\n\nUse this sub-skill when the task is to scale TensorFlow Datasets (TFDS) dataset\npreparation or reading with Apache Beam, cloud runners, file formats, sharding,\nGCS data directories, or `tf.data` performance controls.\n\nRoute other work away:\n\n- Generic `tfds.load`, split syntax, metadata inspection, NumPy\u002FDataFrame, and\n  non-Beam loading recipes -> `data-loading`.\n- Full `tfds` CLI syntax outside Beam\u002Ffile-format\u002Fsharding flags ->\n  `cli-workflows`.\n- Creating ordinary `GeneratorBasedBuilder` datasets, `_info`, features,\n  checksums, dummy data, and `DatasetBuilderTestCase` -> `dataset-authoring`.\n- Folder datasets, external TFRecord layouts, HuggingFace\u002FCroissant\u002Fcommunity\n  catalogs, and dataset collections -> `formats-and-community`.\n\n## Evidence-backed API surface\n\nThe TFDS code and environment verification for this skill established these\nruntime facts:\n\n- Beam generation is configured through `tfds.download.DownloadConfig` fields\n  including `beam_runner`, `beam_options`, `max_examples_per_split`,\n  `num_shards`, `min_shard_size`, `max_shard_size`, and\n  `nondeterministic_order`.\n- `tfds build --beam_pipeline_options=...` stores a comma-separated options\n  string; TFDS turns each `key=value` segment into a Beam `PipelineOptions` flag\n  without launching Beam during parsing.\n- Supported prepared-file format names are `tfrecord`, `array_record`,\n  `riegeli`, and `parquet`; TFRecord is the default. Some formats require extra\n  reader\u002Fwriter dependencies or different reading APIs.\n- `tfds.core.BeamBasedBuilder` still exists but is deprecated; current Beam\n  authoring should normally use `GeneratorBasedBuilder` whose\n  `_generate_examples` returns a Beam `PTransform`\u002F`PCollection`.\n- `tfds.beam.ReadFromTFDS` reads an already prepared TFDS dataset inside a Beam\n  pipeline, parallelizing over dataset shards.\n- Dataflow, Flink, private GCS, and service-account credentials are external\n  services; do not run or mutate them without explicit user approval, budget,\n  and credential boundaries.\n\n## Fast workflow chooser\n\n| User intent | Use |\n|---|---|\n| \"Build this huge TFDS dataset locally just to test the builder\" | Direct\u002Flocal Beam runner plus a tiny `--max_examples_per_split` or programmatic `DownloadConfig(max_examples_per_split=...)`; warn that full local builds can exceed RAM\u002Fdisk\u002Ftime. |\n| \"Run TFDS generation on Dataflow\" | Build a GCS-backed `data_dir`, `staging_location`, `temp_location`, `requirements_file`, and `runner=DataflowRunner`; verify project\u002Fbucket\u002Fcredentials before launch. |\n| \"Use Flink\" | Pass `runner=FlinkRunner` plus Flink version\u002Fconfig-dir options; verify Beam\u002FFlink version compatibility externally. |\n| \"Use TFDS as source inside Beam\" | Use `tfds.beam.ReadFromTFDS(builder, split=...)` only after the dataset is prepared; respect subsplit\u002Fbatch limitations. |\n| \"Improve reading throughput\u002FRAM\" | Tune `ReadConfig`, shuffling, cache\u002Fbatch\u002Fprefetch order, decode\u002Fskipped features, file format, and GCS locality. |\n| \"Change shard count\u002Fsize or file format\" | Use generation flags or `DownloadConfig`\u002F`download_and_prepare(file_format=...)`; validate reader compatibility before rebuilding. |\n\n## Standard playbooks\n\n1. **Confirm the generation target.** For multi-config builders, require the\n   exact dataset\u002Fconfig\u002Fversion. Avoid accidental \"all configs\" builds.\n2. **Choose runner and storage.** Direct runner is for tiny local tests. Use a\n   distributed runner for terabyte-scale datasets. Dataflow jobs should usually\n   write TFDS data to a `gs:\u002F\u002F...\u002Ftensorflow_datasets` root and use a GCS temp\n   and staging area.\n3. **Build pipeline options safely.** Use\n   [`scripts\u002Fbeam_options_helper.py`](scripts\u002Fbeam_options_helper.py) to create\n   or validate `--beam_pipeline_options` strings and get warnings without\n   launching Beam.\n4. **Pass options through TFDS.** See\n   [`references\u002Fbeam-workflows.md`](references\u002Fbeam-workflows.md) for CLI and\n   programmatic examples.\n5. **Set deterministic-order policy.** Default Beam writing preserves TFDS's\n   deterministic ordering contract. Use `nondeterministic_order=True` or\n   `--nondeterministic_order` only when faster unordered output is acceptable\n   and downstream code never depends on example order.\n6. **Control storage format and shards.** See\n   [`references\u002Fperformance-and-storage.md`](references\u002Fperformance-and-storage.md)\n   for `--file_format`, `--num_shards`, `--max_shard_size_mb`, and\n   `DownloadConfig` equivalents.\n7. **Tune readers after generation.** Prefer `shuffle_files=True` for large\n   training datasets, tune `ReadConfig` for distributed workers and memory, skip\n   unused or expensive features, and benchmark the final pipeline.\n8. **Triage failures before rerunning expensive jobs.** Use\n   [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) for Beam,\n   GCS, sharding, determinism, and performance symptoms.\n\n## Safe command skeletons\n\nLocal smoke only:\n\n```bash\ntfds build DATASET[\u002FCONFIG] \\\n  --max_examples_per_split=10 \\\n  --data_dir=\u002Ftmp\u002Ftfds-smoke\n```\n\nDataflow option construction without launching a job:\n\n```bash\npython scripts\u002Fbeam_options_helper.py \\\n  --runner DataflowRunner \\\n  --dataset DATASET \\\n  --project GCP_PROJECT \\\n  --gcs-bucket gs:\u002F\u002FBUCKET \\\n  --requirements-file \u002Ftmp\u002Fbeam_requirements.txt \\\n  --print-build-command\n```\n\nProgrammatic generation:\n\n```python\nimport apache_beam as beam\nimport tensorflow_datasets as tfds\n\nflags = [\n    \"--runner=DataflowRunner\",\n    \"--project=GCP_PROJECT\",\n    \"--staging_location=gs:\u002F\u002FBUCKET\u002Fbinaries\",\n    \"--temp_location=gs:\u002F\u002FBUCKET\u002Ftemp\",\n    \"--requirements_file=\u002Ftmp\u002Fbeam_requirements.txt\",\n]\n\ndownload_config = tfds.download.DownloadConfig(\n    beam_options=beam.options.pipeline_options.PipelineOptions(flags=flags),\n    # Only set when unordered output is acceptable:\n    # nondeterministic_order=True,\n)\nbuilder = tfds.builder(\"DATASET\u002FCONFIG\", data_dir=\"gs:\u002F\u002FBUCKET\u002Ftensorflow_datasets\")\nbuilder.download_and_prepare(download_config=download_config)\n```\n\n## Verification expectations\n\nFor this sub-skill, safe verification should avoid full downloads and cloud job\nlaunches unless the user explicitly grants budget and credentials. Prefer:\n\n- `python scripts\u002Fbeam_options_helper.py --help` and representative option\n  construction checks.\n- Import\u002Fsignature checks for `tfds.download.DownloadConfig`, `tfds.ReadConfig`,\n  and `tfds.beam.ReadFromTFDS` in an environment that has the required optional\n  packages.\n- Tiny local Beam\u002Fdummy-data checks only when Apache Beam and TensorFlow are\n  already installed.\n- Documentation of skipped Dataflow\u002FFlink\u002Fprivate-GCS checks as external-service\n  gaps, not failures of local validation.\n",{"name":5,"license":23,"metadata":24,"description":7,"disable-model-invocation":26},"Apache 2.0",{"disco-role":25},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-datasets\u002Fsub-skills\u002Fbeam-and-performance",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":26},[36,39,42],{"tool":37,"command":38},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill beam-and-performance",{"tool":40,"command":41},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill beam-and-performance",{"tool":43,"command":44},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill beam-and-performance","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-datasets\u002Fsub-skills\u002Fbeam-and-performance\u002FSKILL.md",[47,53,57,61],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Fbeam-workflows.md","REFERENCE","参考文档",7803,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-datasets\u002Fsub-skills\u002Fbeam-and-performance\u002Freferences\u002Fbeam-workflows.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"references\u002Fperformance-and-storage.md",7956,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-datasets\u002Fsub-skills\u002Fbeam-and-performance\u002Freferences\u002Fperformance-and-storage.md",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Ftroubleshooting.md",8464,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-datasets\u002Fsub-skills\u002Fbeam-and-performance\u002Freferences\u002Ftroubleshooting.md",{"path":62,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":65,"sourceHtmlUrl":66},"scripts\u002Fbeam_options_helper.py","SCRIPT","脚本",16430,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftensorflow-datasets\u002Fsub-skills\u002Fbeam-and-performance\u002Fscripts\u002Fbeam_options_helper.py",[68,75,82,89,96,103],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"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.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"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":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"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":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"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":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"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":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"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",[],[]]