[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M790H5NT45R9KXRA3CSKE3J4":3,"skill-same:01M790H5NT45R9KXRA3CSKE3J4":74},{"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},"01M790H5NT45R9KXRA3CSKE3J4","mmaudio",null,"Route MMAudio video-to-audio, text-to-audio, data preparation, CUDA training, and evaluation workflows.","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\u002Fmmaudio\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: mmaudio\ndescription: \"Route MMAudio video-to-audio, text-to-audio, data preparation,\n  CUDA training, and evaluation workflows.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: MIT\n---\n\n# MMAudio\n\nUse this repo skill when a task involves **MMAudio**, the PyTorch project for\nsynchronized audio generation from video and\u002For text. It is an operating guide\nfor future agents: it distills the repository's install, model assets, CLI\u002FAPI,\ntraining, data-preparation, and evaluation behavior without requiring the agent\nto reopen the original source tree for normal use.\n\n## First checks\n\n1. Confirm the task is about MMAudio-style audio generation, video-to-audio,\n   text-to-audio, the MMAudio model variants, training feature memmaps, or the\n   repo's batch-evaluation\u002Fonset tooling.\n2. Check whether the user needs a real model run. Real inference may download\n   multi-GB model assets and is GPU\u002FMPS-preferred; training\u002Fevaluation paths are\n   CUDA\u002FDDP-oriented.\n3. If a checkout or package install is present, run\n   [`scripts\u002Fcheck_mmaudio_env.py`](scripts\u002Fcheck_mmaudio_env.py) for a safe\n   import\u002Fbackend\u002Fconfig sanity check. It does not download weights or run a\n   model.\n4. Read [`references\u002Frepo-provenance.md`](references\u002Frepo-provenance.md) before\n   deciding whether this skill is stale for a newer checkout.\n\n## Route map\n\n| User intent | Read this |\n| --- | --- |\n| Generate audio for one prompt\u002Fvideo, build a `demo.py` command, use Gradio, or call `mmaudio.eval_utils.generate` | [`sub-skills\u002Finference\u002FSKILL.md`](sub-skills\u002Finference\u002FSKILL.md) |\n| Partition audio, validate captions\u002Fclip manifests, plan CLIP\u002FSynchformer\u002FVAE feature extraction, or inspect TensorDict memmap schemas | [`sub-skills\u002Fdata-preparation\u002FSKILL.md`](sub-skills\u002Fdata-preparation\u002FSKILL.md) |\n| Launch or debug DDP training, bounded smoke runs, checkpoint\u002Fweights resume, EMA synthesis, or Hydra training overrides | [`sub-skills\u002Ftraining\u002FSKILL.md`](sub-skills\u002Ftraining\u002FSKILL.md) |\n| Run batched generation for AudioCaps\u002FVGGSound\u002FMovieGen-style evaluation data or compute onset metrics | [`sub-skills\u002Fevaluation\u002FSKILL.md`](sub-skills\u002Fevaluation\u002FSKILL.md) |\n\n## Shared references\n\n- [`references\u002Fmodel-assets.md`](references\u002Fmodel-assets.md) lists MMAudio model\n  variants, required checkpoint filenames, download\u002Fchecksum behavior, sample\n  rates, and license caveats.\n- [`references\u002Fconfiguration.md`](references\u002Fconfiguration.md) summarizes shared\n  Hydra conventions, sequence-length facts, default paths, and override style.\n- [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) covers\n  cross-cutting install\u002Fimport, CUDA, download, media, and dependency failures.\n- [`references\u002Frepo-routing-metadata.json`](references\u002Frepo-routing-metadata.json)\n  is structured metadata for managed repo-skill routing.\n\n## Installation and environment baseline\n\nMMAudio is a Python package named `mmaudio` and requires Python 3.9+. The\nrepository documentation recommends a Miniforge\u002FConda-style environment on\nUbuntu. Install a CUDA-capable PyTorch stack first when GPU workflows are in\nscope, then install the package in editable mode:\n\n```bash\npython -m pip install torch torchvision torchaudio --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu118 --upgrade\npython -m pip install -e \u003CMMAudio checkout>\n```\n\nThe package metadata declares dependencies such as `torchcodec`,\n`huggingface_hub`, `hydra-core`, `gradio`, `open_clip_torch`, `av`, `timm`,\n`tensordict`, `librosa`, `torchdiffeq`, and `nitrous-ema`. Training and the\nbuilt-in metric\u002Fsample path also depend on the external `av_bench` package from\nthe MMAudio authors' av-benchmark project.\n\n## Model families at a glance\n\n| Variant | Mode | Typical use |\n| --- | --- | --- |\n| `small_16k` | 16 kHz | Lower sample rate, smaller network, training\u002Fevaluation examples. |\n| `small_44k` | 44.1 kHz | Smaller high-sample-rate generation. |\n| `medium_44k` | 44.1 kHz | Larger high-sample-rate model. |\n| `large_44k` | 44.1 kHz | Heavy high-sample-rate model. |\n| `large_44k_v2` | 44.1 kHz | Recommended default for general inference. |\n\nEvery inference\u002Fevaluation variant needs a flow-prediction checkpoint, a VAE,\nand the Synchformer checkpoint; 16 kHz additionally uses the BigVGAN vocoder\nfile. See the model-assets reference before downloading or debugging weights.\n\n## Boundaries and cautions\n\n- This skill is for operating MMAudio, not for general audio ML, TTS, ASR, or\n  unrelated diffusion models.\n- Full generation, training, and batch evaluation can be expensive and may\n  download large files. Use the bundled command builders and schema checks\n  before launching CUDA work.\n- Do not treat CPU import success as proof that CUDA training, feature\n  extraction, or batch evaluation will work. Those code paths use CUDA\u002FNCCL.\n- The generated helper scripts print or validate commands by default; they do\n  not modify model weights, launch servers, download checkpoints, or start\n  training unless a user separately runs the printed project command.\n","# MMAudio\n\nUse this repo skill when a task involves **MMAudio**, the PyTorch project for\nsynchronized audio generation from video and\u002For text. It is an operating guide\nfor future agents: it distills the repository's install, model assets, CLI\u002FAPI,\ntraining, data-preparation, and evaluation behavior without requiring the agent\nto reopen the original source tree for normal use.\n\n## First checks\n\n1. Confirm the task is about MMAudio-style audio generation, video-to-audio,\n   text-to-audio, the MMAudio model variants, training feature memmaps, or the\n   repo's batch-evaluation\u002Fonset tooling.\n2. Check whether the user needs a real model run. Real inference may download\n   multi-GB model assets and is GPU\u002FMPS-preferred; training\u002Fevaluation paths are\n   CUDA\u002FDDP-oriented.\n3. If a checkout or package install is present, run\n   [`scripts\u002Fcheck_mmaudio_env.py`](scripts\u002Fcheck_mmaudio_env.py) for a safe\n   import\u002Fbackend\u002Fconfig sanity check. It does not download weights or run a\n   model.\n4. Read [`references\u002Frepo-provenance.md`](references\u002Frepo-provenance.md) before\n   deciding whether this skill is stale for a newer checkout.\n\n## Route map\n\n| User intent | Read this |\n| --- | --- |\n| Generate audio for one prompt\u002Fvideo, build a `demo.py` command, use Gradio, or call `mmaudio.eval_utils.generate` | [`sub-skills\u002Finference\u002FSKILL.md`](sub-skills\u002Finference\u002FSKILL.md) |\n| Partition audio, validate captions\u002Fclip manifests, plan CLIP\u002FSynchformer\u002FVAE feature extraction, or inspect TensorDict memmap schemas | [`sub-skills\u002Fdata-preparation\u002FSKILL.md`](sub-skills\u002Fdata-preparation\u002FSKILL.md) |\n| Launch or debug DDP training, bounded smoke runs, checkpoint\u002Fweights resume, EMA synthesis, or Hydra training overrides | [`sub-skills\u002Ftraining\u002FSKILL.md`](sub-skills\u002Ftraining\u002FSKILL.md) |\n| Run batched generation for AudioCaps\u002FVGGSound\u002FMovieGen-style evaluation data or compute onset metrics | [`sub-skills\u002Fevaluation\u002FSKILL.md`](sub-skills\u002Fevaluation\u002FSKILL.md) |\n\n## Shared references\n\n- [`references\u002Fmodel-assets.md`](references\u002Fmodel-assets.md) lists MMAudio model\n  variants, required checkpoint filenames, download\u002Fchecksum behavior, sample\n  rates, and license caveats.\n- [`references\u002Fconfiguration.md`](references\u002Fconfiguration.md) summarizes shared\n  Hydra conventions, sequence-length facts, default paths, and override style.\n- [`references\u002Ftroubleshooting.md`](references\u002Ftroubleshooting.md) covers\n  cross-cutting install\u002Fimport, CUDA, download, media, and dependency failures.\n- [`references\u002Frepo-routing-metadata.json`](references\u002Frepo-routing-metadata.json)\n  is structured metadata for managed repo-skill routing.\n\n## Installation and environment baseline\n\nMMAudio is a Python package named `mmaudio` and requires Python 3.9+. The\nrepository documentation recommends a Miniforge\u002FConda-style environment on\nUbuntu. Install a CUDA-capable PyTorch stack first when GPU workflows are in\nscope, then install the package in editable mode:\n\n```bash\npython -m pip install torch torchvision torchaudio --index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu118 --upgrade\npython -m pip install -e \u003CMMAudio checkout>\n```\n\nThe package metadata declares dependencies such as `torchcodec`,\n`huggingface_hub`, `hydra-core`, `gradio`, `open_clip_torch`, `av`, `timm`,\n`tensordict`, `librosa`, `torchdiffeq`, and `nitrous-ema`. Training and the\nbuilt-in metric\u002Fsample path also depend on the external `av_bench` package from\nthe MMAudio authors' av-benchmark project.\n\n## Model families at a glance\n\n| Variant | Mode | Typical use |\n| --- | --- | --- |\n| `small_16k` | 16 kHz | Lower sample rate, smaller network, training\u002Fevaluation examples. |\n| `small_44k` | 44.1 kHz | Smaller high-sample-rate generation. |\n| `medium_44k` | 44.1 kHz | Larger high-sample-rate model. |\n| `large_44k` | 44.1 kHz | Heavy high-sample-rate model. |\n| `large_44k_v2` | 44.1 kHz | Recommended default for general inference. |\n\nEvery inference\u002Fevaluation variant needs a flow-prediction checkpoint, a VAE,\nand the Synchformer checkpoint; 16 kHz additionally uses the BigVGAN vocoder\nfile. See the model-assets reference before downloading or debugging weights.\n\n## Boundaries and cautions\n\n- This skill is for operating MMAudio, not for general audio ML, TTS, ASR, or\n  unrelated diffusion models.\n- Full generation, training, and batch evaluation can be expensive and may\n  download large files. Use the bundled command builders and schema checks\n  before launching CUDA work.\n- Do not treat CPU import success as proof that CUDA training, feature\n  extraction, or batch evaluation will work. Those code paths use CUDA\u002FNCCL.\n- The generated helper scripts print or validate commands by default; they do\n  not modify model weights, launch servers, download checkpoints, or start\n  training unless a user separately runs the printed project command.\n",{"name":5,"license":14,"metadata":23,"description":7,"disable-model-invocation":25},{"disco-role":24},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fmmaudio",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 mmaudio",{"tool":39,"command":40},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill mmaudio",{"tool":42,"command":43},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill mmaudio","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fmmaudio\u002FSKILL.md",[46,52,56,60,64,68],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fconfiguration.md","REFERENCE","参考文档",4948,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fmmaudio\u002Freferences\u002Fconfiguration.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Fmodel-assets.md",3628,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fmmaudio\u002Freferences\u002Fmodel-assets.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Frepo-provenance.md",2677,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fmmaudio\u002Freferences\u002Frepo-provenance.md",{"path":61,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":62,"sourceHtmlUrl":63},"references\u002Frepo-routing-metadata.json",504,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fmmaudio\u002Freferences\u002Frepo-routing-metadata.json",{"path":65,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":66,"sourceHtmlUrl":67},"references\u002Ftroubleshooting.md",4984,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fmmaudio\u002Freferences\u002Ftroubleshooting.md",{"path":69,"resourceType":70,"resourceTypeDesc":71,"sizeBytes":72,"sourceHtmlUrl":73},"scripts\u002Fcheck_mmaudio_env.py","SCRIPT","脚本",7622,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fmmaudio\u002Fscripts\u002Fcheck_mmaudio_env.py",[75,83,90,97,104,111],{"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":79,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"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":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":79,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"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":91,"name":92,"nameZh":6,"description":93,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":79,"licenseFileUrl":6,"sourceHtmlUrl":94,"categories":95,"keywords":96,"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":98,"name":99,"nameZh":6,"description":100,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":79,"licenseFileUrl":6,"sourceHtmlUrl":101,"categories":102,"keywords":103,"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":105,"name":106,"nameZh":6,"description":107,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":79,"licenseFileUrl":6,"sourceHtmlUrl":108,"categories":109,"keywords":110,"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":112,"name":113,"nameZh":6,"description":114,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":79,"licenseFileUrl":6,"sourceHtmlUrl":115,"categories":116,"keywords":117,"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",[],[]]