[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78VCVTH6QT4GX7J14XTHTZW":3,"skill-same:01M78VCVTH6QT4GX7J14XTHTZW":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},"01M78VCVTH6QT4GX7J14XTHTZW","dalle-pytorch",null,"Operate lucidrains DALLE-pytorch for DALL-E style VAE training, transformer training, generation, CLIP ranking, and backend troubleshooting.","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\u002Fdalle-pytorch\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: dalle-pytorch\ndescription: \"Operate lucidrains DALLE-pytorch for DALL-E style VAE training,\n  transformer training, generation, CLIP ranking, and backend troubleshooting.\"\nmetadata:\n  disco-role: operating\ndisable-model-invocation: true\nlicense: MIT\n---\n\n# DALLE-pytorch operating skill\n\nUse this skill when the task is about the `dalle-pytorch` \u002F DALLE-pytorch package: DALL-E style text-to-image transformer modeling, discrete VAE codebooks, CLIP ranking, image\u002Ftext training data, generation checkpoints, sparse attention, DeepSpeed, Horovod, Apex, or the package's training\u002Fgeneration helper surfaces.\n\n## First checks\n\n1. Confirm the user wants DALLE-pytorch, not DALL-E 2 or a diffusion package.\n2. Check whether they are using a pip install, a local checkout, or only asking for API guidance. The published package exports `DiscreteVAE`, `DALLE`, `CLIP`, `OpenAIDiscreteVAE`, and `VQGanVAE`; the historical training\u002Fgeneration helpers were top-level repo scripts, not package console entry points.\n3. Run the bundled install\u002FAPI smoke when behavior matters:\n\n```bash\npython scripts\u002Fcheck_dalle_pytorch_install.py\n```\n\nRead `references\u002Fapi-reference.md` for public signatures and object relationships. Read `references\u002Ftroubleshooting.md` before fixing install\u002Fimport, torch, VAE download, checkpoint, or backend failures.\n\n## Route by task\n\n| User task | Read next | Why |\n| --- | --- | --- |\n| Train or debug a discrete VAE, choose OpenAI\u002FVQGAN\u002Fcustom VAE, inspect VAE checkpoint contents, or create a VAE command | `sub-skills\u002Fvae-training\u002FSKILL.md` | Owns `DiscreteVAE`, VAE checkpoints, image-folder training, VAE alternatives, and VAE-specific failures. |\n| Train\u002Fresume the DALL-E transformer, validate image-text folders, use WebDataset shards, choose tokenizer flags, or understand DALL-E checkpoint payloads | `sub-skills\u002Fdalle-training\u002FSKILL.md` | Owns transformer training, data layout, tokenizer choices, checkpoint\u002Fresume paths, and data validators. |\n| Generate images\u002Ftext, split prompts, prime from an image crop, use CLIP ranking, or debug generation checkpoints | `sub-skills\u002Fgeneration-and-ranking\u002FSKILL.md` | Owns `generate_images`, `generate_texts`, `CLIP`, prompt\u002Foutput layout, and generation troubleshooting. |\n| Decide on DeepSpeed\u002FHorovod\u002FApex\u002FDocker\u002FCUDA, sparse attention, `--distributed_backend`, or optional backend installation | `sub-skills\u002Fdistributed-and-backends\u002FSKILL.md` | Owns distributed wrappers, backend availability checks, attention variants, Docker and source-build caveats. |\n\n## Safe operating boundaries\n\n- Treat full training and generation as GPU\u002Fside-effect workflows. The original helpers call `.cuda()`, write checkpoints or outputs, and may log to W&B. Use bundled command builders and validators first; run long jobs only after the user approves compute, data, checkpoint, and logging side effects.\n- Do not instantiate `OpenAIDiscreteVAE` blindly on modern torch. The source asserts `torch \u003C= 1.10`; with newer torch the class fails before downloading weights.\n- Do not trigger default OpenAI VAE or VQGAN downloads unless the user wants model-cache\u002Fnetwork side effects.\n- DeepSpeed sparse attention, Apex AMP, Horovod, and Docker GPU builds require machine-specific CUDA\u002Ftoolchain decisions. Route those to `distributed-and-backends` instead of installing broad extras.\n- When a user has only a pip install, prefer API recipes and bundled validators. When they have a checkout containing the historical scripts, bundled command builders can produce matching commands without requiring this skill to link back to source files.\n\n## Quick package smoke\n\nUse this for a fast sanity check:\n\n```bash\npython scripts\u002Fcheck_dalle_pytorch_install.py --include-cuda\n```\n\nExpected high-level signal:\n\n- package metadata and `dalle_pytorch` import succeed;\n- `DiscreteVAE`, `DALLE`, `CLIP`, tokenizer, and data helper signatures are visible;\n- a tiny CPU `DiscreteVAE`\u002F`DALLE`\u002F`CLIP` smoke passes;\n- optional CUDA availability is reported without running training.\n\n## Key facts to remember\n\n- Distribution name: `dalle-pytorch`; import root: `dalle_pytorch`; version baseline: `1.6.6`.\n- `DiscreteVAE` can be trained directly or loaded as a checkpoint payload containing `hparams` and `weights`.\n- `DALLE` freezes the VAE, predicts text and image tokens, and can generate images autoregressively with optional classifier-free `cond_scale`.\n- Image-text folder training pairs image files and `.txt` files by matching stem recursively; each text file may contain multiple newline-separated captions.\n- WebDataset training expects `--wds \u003Cimage_key>,\u003Ccaption_key>` and an image-text folder argument that points to a tar, shard folder, URL brace pattern, or GCS path.\n- CLI helpers are checkout scripts. This skill bundles safe command builders and API smokes instead of copying long CUDA\u002FW&B training scripts.\n\n## References\n\n- `references\u002Fapi-reference.md`: public signatures, outputs, and model relationships.\n- `references\u002Ftroubleshooting.md`: cross-cutting install\u002Fimport\u002Fbackend\u002Fcheckpoint issues.\n- `references\u002Frepo-provenance.md`: source commit and evidence paths for refresh decisions.\n- `references\u002Frepo-routing-metadata.json`: structured router metadata for managed imports.\n","# DALLE-pytorch operating skill\n\nUse this skill when the task is about the `dalle-pytorch` \u002F DALLE-pytorch package: DALL-E style text-to-image transformer modeling, discrete VAE codebooks, CLIP ranking, image\u002Ftext training data, generation checkpoints, sparse attention, DeepSpeed, Horovod, Apex, or the package's training\u002Fgeneration helper surfaces.\n\n## First checks\n\n1. Confirm the user wants DALLE-pytorch, not DALL-E 2 or a diffusion package.\n2. Check whether they are using a pip install, a local checkout, or only asking for API guidance. The published package exports `DiscreteVAE`, `DALLE`, `CLIP`, `OpenAIDiscreteVAE`, and `VQGanVAE`; the historical training\u002Fgeneration helpers were top-level repo scripts, not package console entry points.\n3. Run the bundled install\u002FAPI smoke when behavior matters:\n\n```bash\npython scripts\u002Fcheck_dalle_pytorch_install.py\n```\n\nRead `references\u002Fapi-reference.md` for public signatures and object relationships. Read `references\u002Ftroubleshooting.md` before fixing install\u002Fimport, torch, VAE download, checkpoint, or backend failures.\n\n## Route by task\n\n| User task | Read next | Why |\n| --- | --- | --- |\n| Train or debug a discrete VAE, choose OpenAI\u002FVQGAN\u002Fcustom VAE, inspect VAE checkpoint contents, or create a VAE command | `sub-skills\u002Fvae-training\u002FSKILL.md` | Owns `DiscreteVAE`, VAE checkpoints, image-folder training, VAE alternatives, and VAE-specific failures. |\n| Train\u002Fresume the DALL-E transformer, validate image-text folders, use WebDataset shards, choose tokenizer flags, or understand DALL-E checkpoint payloads | `sub-skills\u002Fdalle-training\u002FSKILL.md` | Owns transformer training, data layout, tokenizer choices, checkpoint\u002Fresume paths, and data validators. |\n| Generate images\u002Ftext, split prompts, prime from an image crop, use CLIP ranking, or debug generation checkpoints | `sub-skills\u002Fgeneration-and-ranking\u002FSKILL.md` | Owns `generate_images`, `generate_texts`, `CLIP`, prompt\u002Foutput layout, and generation troubleshooting. |\n| Decide on DeepSpeed\u002FHorovod\u002FApex\u002FDocker\u002FCUDA, sparse attention, `--distributed_backend`, or optional backend installation | `sub-skills\u002Fdistributed-and-backends\u002FSKILL.md` | Owns distributed wrappers, backend availability checks, attention variants, Docker and source-build caveats. |\n\n## Safe operating boundaries\n\n- Treat full training and generation as GPU\u002Fside-effect workflows. The original helpers call `.cuda()`, write checkpoints or outputs, and may log to W&B. Use bundled command builders and validators first; run long jobs only after the user approves compute, data, checkpoint, and logging side effects.\n- Do not instantiate `OpenAIDiscreteVAE` blindly on modern torch. The source asserts `torch \u003C= 1.10`; with newer torch the class fails before downloading weights.\n- Do not trigger default OpenAI VAE or VQGAN downloads unless the user wants model-cache\u002Fnetwork side effects.\n- DeepSpeed sparse attention, Apex AMP, Horovod, and Docker GPU builds require machine-specific CUDA\u002Ftoolchain decisions. Route those to `distributed-and-backends` instead of installing broad extras.\n- When a user has only a pip install, prefer API recipes and bundled validators. When they have a checkout containing the historical scripts, bundled command builders can produce matching commands without requiring this skill to link back to source files.\n\n## Quick package smoke\n\nUse this for a fast sanity check:\n\n```bash\npython scripts\u002Fcheck_dalle_pytorch_install.py --include-cuda\n```\n\nExpected high-level signal:\n\n- package metadata and `dalle_pytorch` import succeed;\n- `DiscreteVAE`, `DALLE`, `CLIP`, tokenizer, and data helper signatures are visible;\n- a tiny CPU `DiscreteVAE`\u002F`DALLE`\u002F`CLIP` smoke passes;\n- optional CUDA availability is reported without running training.\n\n## Key facts to remember\n\n- Distribution name: `dalle-pytorch`; import root: `dalle_pytorch`; version baseline: `1.6.6`.\n- `DiscreteVAE` can be trained directly or loaded as a checkpoint payload containing `hparams` and `weights`.\n- `DALLE` freezes the VAE, predicts text and image tokens, and can generate images autoregressively with optional classifier-free `cond_scale`.\n- Image-text folder training pairs image files and `.txt` files by matching stem recursively; each text file may contain multiple newline-separated captions.\n- WebDataset training expects `--wds \u003Cimage_key>,\u003Ccaption_key>` and an image-text folder argument that points to a tar, shard folder, URL brace pattern, or GCS path.\n- CLI helpers are checkout scripts. This skill bundles safe command builders and API smokes instead of copying long CUDA\u002FW&B training scripts.\n\n## References\n\n- `references\u002Fapi-reference.md`: public signatures, outputs, and model relationships.\n- `references\u002Ftroubleshooting.md`: cross-cutting install\u002Fimport\u002Fbackend\u002Fcheckpoint issues.\n- `references\u002Frepo-provenance.md`: source commit and evidence paths for refresh decisions.\n- `references\u002Frepo-routing-metadata.json`: structured router metadata for managed imports.\n",{"name":5,"license":14,"metadata":23,"description":7,"disable-model-invocation":25},{"disco-role":24},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fdalle-pytorch",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 dalle-pytorch",{"tool":39,"command":40},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill dalle-pytorch",{"tool":42,"command":43},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill dalle-pytorch","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fdalle-pytorch\u002FSKILL.md",[46,52,56,60,64],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fapi-reference.md","REFERENCE","参考文档",6946,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fdalle-pytorch\u002Freferences\u002Fapi-reference.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Frepo-provenance.md",1557,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fdalle-pytorch\u002Freferences\u002Frepo-provenance.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Frepo-routing-metadata.json",408,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fdalle-pytorch\u002Freferences\u002Frepo-routing-metadata.json",{"path":61,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":62,"sourceHtmlUrl":63},"references\u002Ftroubleshooting.md",4393,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fdalle-pytorch\u002Freferences\u002Ftroubleshooting.md",{"path":65,"resourceType":66,"resourceTypeDesc":67,"sizeBytes":68,"sourceHtmlUrl":69},"scripts\u002Fcheck_dalle_pytorch_install.py","SCRIPT","脚本",4758,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fdalle-pytorch\u002Fscripts\u002Fcheck_dalle_pytorch_install.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",[],[]]