[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M796KGJ8PPDTESB466WAEGXW":3,"skill-same:01M796KGJ8PPDTESB466WAEGXW":71},{"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},"01M796KGJ8PPDTESB466WAEGXW","pytorch",null,"Operate transformer_engine.pytorch for PyTorch layers, precision control, distributed overlap, and debug workflows.","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\u002Ftransformer-engine\u002Fsub-skills\u002Fpytorch\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: pytorch\ndescription: \"Operate transformer_engine.pytorch for PyTorch layers, precision\n  control, distributed overlap, and debug workflows.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: Apache 2.0\n---\n\n# PyTorch\n\nUse this sub-skill when the task is to apply Transformer Engine's PyTorch surface inside a model, training loop, export path, or debugging session.\n\n## Route here for\n\n- Replacing `torch.nn` layers with `transformer_engine.pytorch.Linear`, `GroupedLinear`, `LayerNorm`, `RMSNorm`, `LayerNormLinear`, `LayerNormMLP`, `DotProductAttention`, `MultiheadAttention`, or `TransformerLayer`.\n- BF16\u002FFP16 training or inference with matching `params_dtype` and `torch.autocast`.\n- FP8, MXFP8, or NVFP4 questions, recipe selection, and hardware gating with `te.autocast(...)`.\n- Quantized parameter initialization, quantized tensor inspection, and quantizer choices.\n- Distributed execution, checkpointing, userbuffers, comm\u002FGEMM overlap, CUDA graphs, CPU offload, or op-fuser work.\n- Debugging with Nvidia-DL-Framework-Inspect and Transformer Engine debug features.\n- Import, version, backend, and precision-availability troubleshooting.\n\n## Required references\n\nRead these bundled references before changing code or explaining usage:\n\n1. [`references\u002Fapi-reference.md`](references\u002Fapi-reference.md) for constructors, helpers, tensor classes, and op-fuser surfaces.\n2. [`references\u002Fworkflows.md`](references\u002Fworkflows.md) for minimal BF16\u002FFP16 flows and guarded low-precision recipes.\n3. [`references\u002Fdistributed-and-advanced.md`](references\u002Fdistributed-and-advanced.md) for FSDP, checkpointing, userbuffers, CPU offload, export, and advanced fusion.\n4. [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) for import order, runtime mismatches, and precision failures.\n\n## Smoke helper\n\nRun the bundled smoke script to validate a small BF16 `te.Linear` path without depending on the source repository:\n\n```bash\npython scripts\u002Fpytorch_bf16_smoke.py --help\npython scripts\u002Fpytorch_bf16_smoke.py --device cuda --in-features 16 --out-features 32 --batch-size 4\n```\n\nThe helper prints version, device, and precision-availability facts, then runs a tiny BF16 forward\u002Fbackward pass. It is intentionally narrower than the full `TransformerLayer` examples.\n\n## Operating cautions\n\n- Use `torch.autocast` for BF16\u002FFP16 compute control; use `te.autocast` for FP8\u002FMXFP8\u002FNVFP4 quantized compute. They are not interchangeable.\n- Do not promise FP8 on A100-class hardware. Always check availability first and surface the reason string when a recipe is unavailable.\n- When `quantized_model_init(enabled=True)` is used with master-weight optimizers, preserve or copy high-precision init values before clearing them if you need FP32 seeding.\n- Prefer the bundled smoke path for environment checks when larger TransformerLayer examples are unstable.\n","# PyTorch\n\nUse this sub-skill when the task is to apply Transformer Engine's PyTorch surface inside a model, training loop, export path, or debugging session.\n\n## Route here for\n\n- Replacing `torch.nn` layers with `transformer_engine.pytorch.Linear`, `GroupedLinear`, `LayerNorm`, `RMSNorm`, `LayerNormLinear`, `LayerNormMLP`, `DotProductAttention`, `MultiheadAttention`, or `TransformerLayer`.\n- BF16\u002FFP16 training or inference with matching `params_dtype` and `torch.autocast`.\n- FP8, MXFP8, or NVFP4 questions, recipe selection, and hardware gating with `te.autocast(...)`.\n- Quantized parameter initialization, quantized tensor inspection, and quantizer choices.\n- Distributed execution, checkpointing, userbuffers, comm\u002FGEMM overlap, CUDA graphs, CPU offload, or op-fuser work.\n- Debugging with Nvidia-DL-Framework-Inspect and Transformer Engine debug features.\n- Import, version, backend, and precision-availability troubleshooting.\n\n## Required references\n\nRead these bundled references before changing code or explaining usage:\n\n1. [`references\u002Fapi-reference.md`](references\u002Fapi-reference.md) for constructors, helpers, tensor classes, and op-fuser surfaces.\n2. [`references\u002Fworkflows.md`](references\u002Fworkflows.md) for minimal BF16\u002FFP16 flows and guarded low-precision recipes.\n3. [`references\u002Fdistributed-and-advanced.md`](references\u002Fdistributed-and-advanced.md) for FSDP, checkpointing, userbuffers, CPU offload, export, and advanced fusion.\n4. [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) for import order, runtime mismatches, and precision failures.\n\n## Smoke helper\n\nRun the bundled smoke script to validate a small BF16 `te.Linear` path without depending on the source repository:\n\n```bash\npython scripts\u002Fpytorch_bf16_smoke.py --help\npython scripts\u002Fpytorch_bf16_smoke.py --device cuda --in-features 16 --out-features 32 --batch-size 4\n```\n\nThe helper prints version, device, and precision-availability facts, then runs a tiny BF16 forward\u002Fbackward pass. It is intentionally narrower than the full `TransformerLayer` examples.\n\n## Operating cautions\n\n- Use `torch.autocast` for BF16\u002FFP16 compute control; use `te.autocast` for FP8\u002FMXFP8\u002FNVFP4 quantized compute. They are not interchangeable.\n- Do not promise FP8 on A100-class hardware. Always check availability first and surface the reason string when a recipe is unavailable.\n- When `quantized_model_init(enabled=True)` is used with master-weight optimizers, preserve or copy high-precision init values before clearing them if you need FP32 seeding.\n- Prefer the bundled smoke path for environment checks when larger TransformerLayer examples are unstable.\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\u002Ftransformer-engine\u002Fsub-skills\u002Fpytorch",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 pytorch",{"tool":40,"command":41},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill pytorch",{"tool":43,"command":44},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill pytorch","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftransformer-engine\u002Fsub-skills\u002Fpytorch\u002FSKILL.md",[47,53,57,61,65],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Fapi-reference.md","REFERENCE","参考文档",12466,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftransformer-engine\u002Fsub-skills\u002Fpytorch\u002Freferences\u002Fapi-reference.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"references\u002Fdistributed-and-advanced.md",6689,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftransformer-engine\u002Fsub-skills\u002Fpytorch\u002Freferences\u002Fdistributed-and-advanced.md",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Ftroubleshooting.md",4070,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftransformer-engine\u002Fsub-skills\u002Fpytorch\u002Freferences\u002Ftroubleshooting.md",{"path":62,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":63,"sourceHtmlUrl":64},"references\u002Fworkflows.md",4168,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftransformer-engine\u002Fsub-skills\u002Fpytorch\u002Freferences\u002Fworkflows.md",{"path":66,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":69,"sourceHtmlUrl":70},"scripts\u002Fpytorch_bf16_smoke.py","SCRIPT","脚本",4733,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ftransformer-engine\u002Fsub-skills\u002Fpytorch\u002Fscripts\u002Fpytorch_bf16_smoke.py",[72,79,86,93,100,107],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"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.","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":14,"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":14,"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":14,"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":14,"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":14,"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",[],[]]