[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"extension-skill-huggingface-huggingface-local-models-de":3,"guides-for-huggingface-huggingface-local-models":747,"similar-k170rhxp71d7qg10dfv2hybw1n86n8dm-de":748},{"_creationTime":4,"_id":5,"children":6,"community":7,"display":9,"evaluation":15,"identity":255,"isFallback":252,"parentExtension":260,"providers":294,"relations":298,"repo":299,"tags":745,"workflow":746},1778690773482.488,"k170rhxp71d7qg10dfv2hybw1n86n8dm",[],{"reviewCount":8},0,{"description":10,"installMethods":11,"name":13,"sourceUrl":14},"Use to select models to run locally with llama.cpp and GGUF on CPU, Mac Metal, CUDA, or ROCm. Covers finding GGUFs, quant selection, running servers, exact GGUF file lookup, conversion, and OpenAI-compatible local serving.",{"claudeCode":12},"huggingface/skills","Hugging Face Local Models","https://github.com/huggingface/skills",{"_creationTime":16,"_id":17,"extensionId":5,"locale":18,"result":19,"trustSignals":236,"workflow":253},1778691335553.4485,"kn7ew3rmhfxj6c6bvzywh41xss86n69k","en",{"checks":20,"evaluatedAt":193,"extensionSummary":194,"features":195,"nonGoals":201,"practices":205,"prerequisites":206,"promptVersionExtension":210,"promptVersionScoring":211,"purpose":212,"rationale":213,"score":214,"summary":215,"tags":216,"targetMarket":223,"tier":224,"useCases":225,"workflow":230},[21,26,29,32,36,39,44,49,52,55,59,63,66,70,73,76,79,82,85,88,92,96,100,104,108,111,114,117,121,124,127,130,133,136,139,143,147,151,154,158,161,164,167,170,174,177,180,183,186,190],{"category":22,"check":23,"severity":24,"summary":25},"Practical Utility","Problem relevance","pass","The description clearly states the problem of selecting and running local LLM models with llama.cpp and GGUF, covering finding models, quantization, and serving.",{"category":22,"check":27,"severity":24,"summary":28},"Unique selling proposition","The skill provides a structured workflow and specific guidance for interacting with Hugging Face Hub for local model deployment, going beyond basic model selection.",{"category":22,"check":30,"severity":24,"summary":31},"Production readiness","The skill covers the full lifecycle from model discovery and selection to local serving and conversion, providing practical commands and references.",{"category":33,"check":34,"severity":24,"summary":35},"Scope","Single responsibility principle","The skill focuses on the specific domain of running Hugging Face GGUF models locally with llama.cpp, including related tasks like discovery and conversion.",{"category":33,"check":37,"severity":24,"summary":38},"Description quality","The displayed description accurately reflects the skill's capabilities in selecting and running local GGUF models with llama.cpp.",{"category":40,"check":41,"severity":42,"summary":43},"Invocation","Scoped tools","not_applicable","This skill does not expose discrete tools; its functionality is described via prose and command examples.",{"category":45,"check":46,"severity":47,"summary":48},"Documentation","Configuration & parameter reference","info","While the skill details model selection and launch commands, explicit documentation on configuration parameters for `llama-cli` or `llama-server` beyond basic flags is not provided.",{"category":33,"check":50,"severity":42,"summary":51},"Tool naming","The skill does not expose named tools; it guides the user through commands and workflows.",{"category":33,"check":53,"severity":42,"summary":54},"Minimal I/O surface","The skill does not expose specific tools with parameter schemas or response shapes.",{"category":56,"check":57,"severity":24,"summary":58},"License","License usability","The extension is licensed under the Apache-2.0 license, as indicated by the bundled LICENSE file.",{"category":60,"check":61,"severity":24,"summary":62},"Maintenance","Commit recency","The repository has recent commits, with the latest push on 2026-05-12, indicating active maintenance.",{"category":60,"check":64,"severity":42,"summary":65},"Dependency Management","The skill does not appear to bundle or manage third-party dependencies directly; it relies on the user installing external tools like llama.cpp.",{"category":67,"check":68,"severity":24,"summary":69},"Security","Secret Management","No secrets are used or handled by the skill itself; it instructs the user to authenticate with Hugging Face if necessary.",{"category":67,"check":71,"severity":24,"summary":72},"Injection","The skill focuses on providing commands and guidance, and does not appear to load or execute untrusted external data as instructions.",{"category":67,"check":74,"severity":24,"summary":75},"Transitive Supply-Chain Grenades","The skill instructs users to install `llama.cpp` and `hf` CLI, but does not fetch or execute code from arbitrary remote URLs at runtime.",{"category":67,"check":77,"severity":24,"summary":78},"Sandbox Isolation","The skill provides commands and instructions; it does not modify files outside of the user's control or project folders.",{"category":67,"check":80,"severity":24,"summary":81},"Sandbox escape primitives","No detached-process spawns or deny-retry loops are evident in the provided skill instructions.",{"category":67,"check":83,"severity":24,"summary":84},"Data Exfiltration","The skill does not instruct the agent to read or submit confidential data to third parties.",{"category":67,"check":86,"severity":24,"summary":87},"Hidden Text Tricks","The bundled content and descriptions are free of hidden steering tricks.",{"category":89,"check":90,"severity":24,"summary":91},"Hooks","Opaque code execution","The skill instructions are plain text and do not involve obfuscated code or runtime script fetching.",{"category":93,"check":94,"severity":24,"summary":95},"Portability","Structural Assumption","The skill's instructions are general and do not make assumptions about specific user project directory structures.",{"category":97,"check":98,"severity":24,"summary":99},"Trust","Issues Attention","In the last 90 days, 4 issues were opened and 6 were closed, indicating active engagement from maintainers with a closure rate of >= 50%.",{"category":101,"check":102,"severity":47,"summary":103},"Versioning","Release Management","There is no explicit versioning (e.g., semver in frontmatter, CHANGELOG, or release tags) for this skill; installation instructions reference the `main` branch.",{"category":105,"check":106,"severity":42,"summary":107},"Code Execution","Validation","The skill itself does not contain executable code or structured output that would require schema validation.",{"category":67,"check":109,"severity":24,"summary":110},"Unguarded Destructive Operations","The skill provides instructions for running models and conversion scripts; it does not involve inherently destructive operations.",{"category":105,"check":112,"severity":42,"summary":113},"Error Handling","The skill does not contain its own executable code with error paths; it instructs the user on how to execute external tools.",{"category":105,"check":115,"severity":42,"summary":116},"Logging","The skill does not perform actions that require local audit logging; it guides the execution of external tools.",{"category":118,"check":119,"severity":42,"summary":120},"Compliance","GDPR","The skill operates on model files and CLI commands, not personal data.",{"category":118,"check":122,"severity":24,"summary":123},"Target market","The skill provides instructions for using local models and tools, with no regional or jurisdictional restrictions detected.",{"category":93,"check":125,"severity":24,"summary":126},"Runtime stability","The skill provides instructions for various hardware acceleration methods (Metal, CUDA, ROCm, CPU) and build options for `llama.cpp`.",{"category":45,"check":128,"severity":24,"summary":129},"README","A README.md file exists and provides a comprehensive overview of Hugging Face skills, including installation and contribution guidelines.",{"category":33,"check":131,"severity":42,"summary":132},"Tool surface size","This skill does not expose a list of tools; it is a procedural guide.",{"category":40,"check":134,"severity":42,"summary":135},"Overlapping near-synonym tools","The skill does not expose multiple tools, so there are no overlapping synonyms.",{"category":45,"check":137,"severity":24,"summary":138},"Phantom features","All advertised capabilities, such as finding GGUFs, quant selection, and running servers, are directly supported by the provided instructions and references.",{"category":140,"check":141,"severity":24,"summary":142},"Install","Installation instruction","Installation instructions for `llama.cpp` are provided, along with clear examples for searching the Hub, running models directly, and converting models.",{"category":144,"check":145,"severity":42,"summary":146},"Errors","Actionable error messages","The skill itself does not generate user-facing errors; it relies on the error messages from the underlying tools like `llama.cpp`.",{"category":148,"check":149,"severity":24,"summary":150},"Execution","Pinned dependencies","The skill instructs users to install `llama.cpp` and `hf` CLI, implying standard installation methods which typically pin dependencies.",{"category":33,"check":152,"severity":42,"summary":153},"Dry-run preview","The skill's focus is on model selection and running, not state-changing operations that would benefit from a dry-run.",{"category":155,"check":156,"severity":42,"summary":157},"Protocol","Idempotent retry & timeouts","The skill does not involve remote calls or state-changing operations that would require specific retry or timeout handling.",{"category":118,"check":159,"severity":42,"summary":160},"Telemetry opt-in","The skill does not emit telemetry; it guides local execution.",{"category":40,"check":162,"severity":24,"summary":163},"Precise Purpose","The skill precisely describes its purpose (selecting and running local llama.cpp/GGUF models) and provides clear use cases and boundaries.",{"category":40,"check":165,"severity":24,"summary":166},"Concise Frontmatter","The frontmatter is concise and effectively summarizes the skill's core capability for precise routing.",{"category":45,"check":168,"severity":24,"summary":169},"Concise Body","The main body of the SKILL.md is concise and delegates detailed information to separate reference files.",{"category":171,"check":172,"severity":24,"summary":173},"Context","Progressive Disclosure","The SKILL.md outlines the workflow and links to detailed reference files for quantization, hub discovery, and hardware acceleration.",{"category":171,"check":175,"severity":42,"summary":176},"Forked exploration","This skill does not involve deep exploration or code review; it provides direct instructions for model selection and execution.",{"category":22,"check":178,"severity":24,"summary":179},"Usage examples","Sufficient end-to-end examples are provided for installing llama.cpp, searching the Hub, running models directly, and converting models.",{"category":22,"check":181,"severity":47,"summary":182},"Edge cases","While general guidance on quant choice and troubleshooting is provided, specific failure modes with symptoms and recovery steps are not extensively detailed.",{"category":105,"check":184,"severity":42,"summary":185},"Tool Fallback","This skill relies on external tools (`llama.cpp`, `hf CLI`) and does not have fallbacks for them.",{"category":187,"check":188,"severity":42,"summary":189},"Safety","Halt on unexpected state","The skill does not perform destructive operations or have preconditions that would require halting on unexpected state.",{"category":93,"check":191,"severity":24,"summary":192},"Cross-skill coupling","The skill is self-contained and does not implicitly rely on other skills being loaded.",1778691335061,"This skill guides users through selecting, downloading, and running large language models compatible with llama.cpp and the GGUF format. It covers finding models on Hugging Face Hub, choosing appropriate quantization levels, and executing models via `llama-cli` or `llama-server`, including conversion steps when necessary.",[196,197,198,199,200],"Find GGUF models on Hugging Face Hub","Select optimal quantization levels","Run models with llama-cli and llama-server","Convert models from Transformers to GGUF","Support for CPU, Metal, CUDA, and ROCm",[202,203,204],"Training or fine-tuning models","Managing Hugging Face Hub repositories directly (beyond downloading)","Providing a full GUI for model management",[],[207,208,209],"llama.cpp installed","Python 3","Hugging Face Hub CLI (optional for authentication)","3.0.0","4.4.0","To enable users to easily select and run local language models using llama.cpp and GGUF formats, covering model discovery, quantization, and serving.","The skill is highly practical and well-documented, with clear instructions, examples, and organized references. It addresses a specific user need effectively. Release management is informational due to lack of explicit versioning.",95,"Excellent skill for running local LLMs with llama.cpp, offering comprehensive guidance.",[217,218,219,220,221,222],"llm","local","models","llama-cpp","gguf","huggingface","global","verified",[226,227,228,229],"Selecting the best GGUF model for your hardware","Running LLMs locally for privacy or cost savings","Experimenting with different model quantizations","Setting up an OpenAI-compatible local inference server",[231,232,233,234,235],"Search Hugging Face Hub for llama.cpp-compatible GGUF models.","Identify the recommended quant and file from the model's page or tree API.","Install or ensure `llama.cpp` is available.","Launch the model using `llama-cli` or `llama-server` with appropriate flags.","Convert models from Transformers to GGUF if no pre-quantized version is available.",{"codeQuality":237,"collectedAt":239,"documentation":240,"maintenance":243,"security":249,"testCoverage":251},{"hasLockfile":238},false,1778691314433,{"descriptionLength":241,"readmeSize":242},222,9821,{"closedIssues90d":244,"forks":245,"hasChangelog":238,"openIssues90d":246,"pushedAt":247,"stars":248},6,663,4,1778593131000,10482,{"hasNpmPackage":238,"license":250,"smitheryVerified":238},"Apache-2.0",{"hasCi":252,"hasTests":238},true,{"updatedAt":254},1778691335553,{"basePath":256,"githubOwner":222,"githubRepo":257,"locale":18,"slug":258,"type":259},"skills/huggingface-local-models","skills","huggingface-local-models","skill",{"_creationTime":261,"_id":262,"community":263,"display":264,"identity":269,"parentExtension":272,"providers":273,"relations":288,"tags":290,"workflow":291},1778690773482.486,"k175g1spb5757qt4tnj9cktcn986mshy",{"reviewCount":8},{"description":265,"installMethods":266,"name":268,"sourceUrl":14},"Agent Skills for AI/ML tasks including dataset creation, model training, evaluation, and research paper publishing on Hugging Face Hub",{"claudeCode":267},"huggingface-skills","Hugging Face Skills",{"basePath":270,"githubOwner":222,"githubRepo":257,"locale":18,"slug":257,"type":271},"","plugin",null,{"evaluate":274,"extract":283},{"promptVersionExtension":210,"promptVersionScoring":211,"score":275,"tags":276,"targetMarket":223,"tier":224},98,[222,277,278,279,219,280,281,282],"ai","ml","datasets","training","cli","python",{"commitSha":284,"license":250,"plugin":285},"HEAD",{"mcpCount":8,"provider":286,"skillCount":287},"classify",14,{"repoId":289},"kd72xwt5xnc0ktc4p7smzfcp3986m959",[277,281,279,222,278,219,282,280],{"evaluatedAt":292,"extractAt":293,"updatedAt":292},1778691185872,1778690773482,{"evaluate":295,"extract":297},{"promptVersionExtension":210,"promptVersionScoring":211,"score":214,"tags":296,"targetMarket":223,"tier":224},[217,218,219,220,221,222],{"commitSha":284,"license":250},{"parentExtensionId":262,"repoId":289},{"_creationTime":300,"_id":289,"identity":301,"providers":302,"workflow":741},1778689536128.5474,{"githubOwner":222,"githubRepo":257,"sourceUrl":14},{"classify":303,"discover":734,"github":737},{"commitSha":284,"extensions":304},[305,319,328,333,341,349,357,365,373,381,389,397,405,413,421,429,472,480,486,492,509,515,522,564,575,594,600,620,632,656,714],{"basePath":270,"description":265,"displayName":267,"installMethods":306,"rationale":307,"selectedPaths":308,"source":317,"sourceLanguage":18,"type":318},{"claudeCode":12},"marketplace.json at .claude-plugin/marketplace.json",[309,312,314],{"path":310,"priority":311},".claude-plugin/marketplace.json","mandatory",{"path":313,"priority":311},"README.md",{"path":315,"priority":316},"LICENSE","high","rule","marketplace",{"basePath":320,"description":321,"displayName":322,"installMethods":323,"rationale":324,"selectedPaths":325,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-llm-trainer","Train or fine-tune language models using TRL on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes hardware selection, cost estimation, Trackio monitoring, and Hub persistence.","huggingface-llm-trainer",{"claudeCode":322},"inline plugin source from marketplace.json at skills/huggingface-llm-trainer",[326],{"path":327,"priority":316},"SKILL.md",{"basePath":256,"description":10,"displayName":258,"installMethods":329,"rationale":330,"selectedPaths":331,"source":317,"sourceLanguage":18,"type":271},{"claudeCode":258},"inline plugin source from marketplace.json at skills/huggingface-local-models",[332],{"path":327,"priority":316},{"basePath":334,"description":335,"displayName":336,"installMethods":337,"rationale":338,"selectedPaths":339,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-paper-publisher","Publish and manage research papers on Hugging Face Hub. Supports creating paper pages, linking papers to models/datasets, claiming authorship, and generating professional markdown-based research articles.","huggingface-paper-publisher",{"claudeCode":336},"inline plugin source from marketplace.json at skills/huggingface-paper-publisher",[340],{"path":327,"priority":316},{"basePath":342,"description":343,"displayName":344,"installMethods":345,"rationale":346,"selectedPaths":347,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-papers","Look up and read Hugging Face paper pages in markdown, and use the papers API for structured metadata like authors, linked models, datasets, Spaces, and media URLs when needed.","huggingface-papers",{"claudeCode":344},"inline plugin source from marketplace.json at skills/huggingface-papers",[348],{"path":327,"priority":316},{"basePath":350,"description":351,"displayName":352,"installMethods":353,"rationale":354,"selectedPaths":355,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-community-evals","Add and manage evaluation results in Hugging Face model cards. Supports extracting eval tables from README content, importing scores from Artificial Analysis API, and running custom evaluations with vLLM/lighteval.","huggingface-community-evals",{"claudeCode":352},"inline plugin source from marketplace.json at skills/huggingface-community-evals",[356],{"path":327,"priority":316},{"basePath":358,"description":359,"displayName":360,"installMethods":361,"rationale":362,"selectedPaths":363,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-best","Find the best AI model for any task by querying Hugging Face leaderboards and benchmarks. Recommends top models based on task type, hardware constraints, and benchmark scores.","huggingface-best",{"claudeCode":360},"inline plugin source from marketplace.json at skills/huggingface-best",[364],{"path":327,"priority":316},{"basePath":366,"description":367,"displayName":368,"installMethods":369,"rationale":370,"selectedPaths":371,"source":317,"sourceLanguage":18,"type":271},"skills/hf-cli","Execute Hugging Face Hub operations using the hf CLI. Download models/datasets, upload files, manage repos, and run cloud compute jobs.","hf-cli",{"claudeCode":368},"inline plugin source from marketplace.json at skills/hf-cli",[372],{"path":327,"priority":316},{"basePath":374,"description":375,"displayName":376,"installMethods":377,"rationale":378,"selectedPaths":379,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-trackio","Track and visualize ML training experiments with Trackio. Log metrics via Python API and retrieve them via CLI. Supports real-time dashboards synced to HF Spaces.","huggingface-trackio",{"claudeCode":376},"inline plugin source from marketplace.json at skills/huggingface-trackio",[380],{"path":327,"priority":316},{"basePath":382,"description":383,"displayName":384,"installMethods":385,"rationale":386,"selectedPaths":387,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-datasets","Explore, query, and extract data from any Hugging Face dataset using the Dataset Viewer REST API and npx tooling. Zero Python dependencies — covers split/config discovery, row pagination, text search, filtering, SQL via parquetlens, and dataset upload via CLI.","huggingface-datasets",{"claudeCode":384},"inline plugin source from marketplace.json at skills/huggingface-datasets",[388],{"path":327,"priority":316},{"basePath":390,"description":391,"displayName":392,"installMethods":393,"rationale":394,"selectedPaths":395,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-tool-builder","Build reusable scripts for Hugging Face Hub and API workflows. Useful for chaining API calls, enriching Hub metadata, or automating repeated tasks.","huggingface-tool-builder",{"claudeCode":392},"inline plugin source from marketplace.json at skills/huggingface-tool-builder",[396],{"path":327,"priority":316},{"basePath":398,"description":399,"displayName":400,"installMethods":401,"rationale":402,"selectedPaths":403,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-gradio","Build Gradio web UIs and demos in Python. Use when creating or editing Gradio apps, components, event listeners, layouts, or chatbots.","huggingface-gradio",{"claudeCode":400},"inline plugin source from marketplace.json at skills/huggingface-gradio",[404],{"path":327,"priority":316},{"basePath":406,"description":407,"displayName":408,"installMethods":409,"rationale":410,"selectedPaths":411,"source":317,"sourceLanguage":18,"type":271},"skills/transformers-js","Run state-of-the-art machine learning models directly in JavaScript/TypeScript for NLP, computer vision, audio processing, and multimodal tasks. Works in Node.js and browsers with WebGPU/WASM using Hugging Face models.","transformers-js",{"claudeCode":408},"inline plugin source from marketplace.json at skills/transformers-js",[412],{"path":327,"priority":316},{"basePath":414,"description":415,"displayName":416,"installMethods":417,"rationale":418,"selectedPaths":419,"source":317,"sourceLanguage":18,"type":271},"skills/huggingface-vision-trainer","Train and fine-tune object detection models (RTDETRv2, YOLOS, DETR and others) and image classification models (timm and transformers models — MobileNetV3, MobileViT, ResNet, ViT/DINOv3) using Transformers Trainer API on Hugging Face Jobs infrastructure or locally. Includes COCO dataset format support, Albumentations augmentation, mAP/mAR metrics, trackio tracking, hardware selection, and Hub persistence.","huggingface-vision-trainer",{"claudeCode":416},"inline plugin source from marketplace.json at skills/huggingface-vision-trainer",[420],{"path":327,"priority":316},{"basePath":422,"description":423,"displayName":424,"installMethods":425,"rationale":426,"selectedPaths":427,"source":317,"sourceLanguage":18,"type":271},"skills/train-sentence-transformers","Train or fine-tune sentence-transformers models across all three architectures: SentenceTransformer (bi-encoder embeddings), CrossEncoder (rerankers), and SparseEncoder (SPLADE). Covers loss selection, hard-negative mining, evaluators, distillation, LoRA, Matryoshka, and Hugging Face Hub publishing.","train-sentence-transformers",{"claudeCode":424},"inline plugin source from marketplace.json at skills/train-sentence-transformers",[428],{"path":327,"priority":316},{"basePath":270,"description":265,"displayName":267,"installMethods":430,"license":250,"rationale":431,"selectedPaths":432,"source":317,"sourceLanguage":18,"type":271},{"claudeCode":267},"plugin manifest at .claude-plugin/plugin.json",[433,435,436,437,440,442,444,446,448,450,452,454,456,458,460,462,464,466,468,470],{"path":434,"priority":311},".claude-plugin/plugin.json",{"path":313,"priority":311},{"path":315,"priority":316},{"path":438,"priority":439},"skills/hf-cli/SKILL.md","medium",{"path":441,"priority":439},"skills/huggingface-best/SKILL.md",{"path":443,"priority":439},"skills/huggingface-community-evals/SKILL.md",{"path":445,"priority":439},"skills/huggingface-datasets/SKILL.md",{"path":447,"priority":439},"skills/huggingface-gradio/SKILL.md",{"path":449,"priority":439},"skills/huggingface-llm-trainer/SKILL.md",{"path":451,"priority":439},"skills/huggingface-local-models/SKILL.md",{"path":453,"priority":439},"skills/huggingface-paper-publisher/SKILL.md",{"path":455,"priority":439},"skills/huggingface-papers/SKILL.md",{"path":457,"priority":439},"skills/huggingface-tool-builder/SKILL.md",{"path":459,"priority":439},"skills/huggingface-trackio/SKILL.md",{"path":461,"priority":439},"skills/huggingface-vision-trainer/SKILL.md",{"path":463,"priority":439},"skills/train-sentence-transformers/SKILL.md",{"path":465,"priority":439},"skills/transformers-js/SKILL.md",{"path":467,"priority":311},".mcp.json",{"path":469,"priority":316},"agents/AGENTS.md",{"path":471,"priority":316},".cursor-plugin/plugin.json",{"basePath":473,"description":474,"displayName":475,"installMethods":476,"rationale":477,"selectedPaths":478,"source":317,"sourceLanguage":18,"type":259},"hf-mcp/skills/hf-mcp","Use Hugging Face Hub via MCP server tools. Search models, datasets, Spaces, papers. Get repo details, fetch documentation, run compute jobs, and use Gradio Spaces as AI tools. Available when connected to the HF MCP server.","hf-mcp",{"claudeCode":12},"SKILL.md frontmatter at hf-mcp/skills/hf-mcp/SKILL.md",[479],{"path":327,"priority":311},{"basePath":366,"description":481,"displayName":368,"installMethods":482,"rationale":483,"selectedPaths":484,"source":317,"sourceLanguage":18,"type":259},"Hugging Face Hub CLI (`hf`) for downloading, uploading, and managing models, datasets, spaces, buckets, repos, papers, jobs, and more on the Hugging Face Hub. Use when: handling authentication; managing local cache; managing Hugging Face Buckets; running or scheduling jobs on Hugging Face infrastructure; managing Hugging Face repos; discussions and pull requests; browsing models, datasets and spaces; reading, searching, or browsing academic papers; managing collections; querying datasets; configuring spaces; setting up webhooks; or deploying and managing HF Inference Endpoints. Make sure to use this skill whenever the user mentions 'hf', 'huggingface', 'Hugging Face', 'huggingface-cli', or 'hugging face cli', or wants to do anything related to the Hugging Face ecosystem and to AI and ML in general. Also use for cloud storage needs like training checkpoints, data pipelines, or agent traces. Use even if the user doesn't explicitly ask for a CLI command. Replaces the deprecated `huggingface-cli`.",{"claudeCode":12},"SKILL.md frontmatter at skills/hf-cli/SKILL.md",[485],{"path":327,"priority":311},{"basePath":358,"description":487,"displayName":360,"installMethods":488,"rationale":489,"selectedPaths":490,"source":317,"sourceLanguage":18,"type":259},"Use when the user asks about finding the best, top, or recommended model for a task, wants to know what AI model to use, or wants to compare models by benchmark scores. Triggers on: \"best model for X\", \"what model should I use for\", \"top models for [task]\", \"which model runs on my laptop/machine/device\", \"recommend a model for\", \"what LLM should I use for\", \"compare models for\", \"what's state of the art for\", or any question about choosing an AI model for a specific use case. Always use this skill when the user wants model recommendations or comparisons, even if they don't explicitly mention HuggingFace or benchmarks.\n",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-best/SKILL.md",[491],{"path":327,"priority":311},{"basePath":350,"description":493,"displayName":352,"installMethods":494,"rationale":495,"selectedPaths":496,"source":317,"sourceLanguage":18,"type":259},"Run evaluations for Hugging Face Hub models using inspect-ai and lighteval on local hardware. Use for backend selection, local GPU evals, and choosing between vLLM / Transformers / accelerate. Not for HF Jobs orchestration, model-card PRs, .eval_results publication, or community-evals automation.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-community-evals/SKILL.md",[497,498,501,503,505,507],{"path":327,"priority":311},{"path":499,"priority":500},"examples/.env.example","low",{"path":502,"priority":500},"examples/USAGE_EXAMPLES.md",{"path":504,"priority":500},"scripts/inspect_eval_uv.py",{"path":506,"priority":500},"scripts/inspect_vllm_uv.py",{"path":508,"priority":500},"scripts/lighteval_vllm_uv.py",{"basePath":382,"description":510,"displayName":384,"installMethods":511,"rationale":512,"selectedPaths":513,"source":317,"sourceLanguage":18,"type":259},"Use this skill for Hugging Face Dataset Viewer API workflows that fetch subset/split metadata, paginate rows, search text, apply filters, download parquet URLs, and read size or statistics.\r",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-datasets/SKILL.md",[514],{"path":327,"priority":311},{"basePath":398,"description":399,"displayName":400,"installMethods":516,"rationale":517,"selectedPaths":518,"source":317,"sourceLanguage":18,"type":259},{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-gradio/SKILL.md",[519,520],{"path":327,"priority":311},{"path":521,"priority":439},"examples.md",{"basePath":320,"description":523,"displayName":322,"installMethods":524,"rationale":525,"selectedPaths":526,"source":317,"sourceLanguage":18,"type":259},"Train or fine-tune language and vision models using TRL (Transformer Reinforcement Learning) or Unsloth with Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, model selection/leaderboards and model persistence. Use for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-llm-trainer/SKILL.md",[527,528,530,532,534,536,538,540,542,544,546,548,550,552,554,556,558,560,562],{"path":327,"priority":311},{"path":529,"priority":439},"references/gguf_conversion.md",{"path":531,"priority":439},"references/hardware_guide.md",{"path":533,"priority":439},"references/hub_saving.md",{"path":535,"priority":439},"references/local_training_macos.md",{"path":537,"priority":439},"references/reliability_principles.md",{"path":539,"priority":439},"references/trackio_guide.md",{"path":541,"priority":439},"references/training_methods.md",{"path":543,"priority":439},"references/training_patterns.md",{"path":545,"priority":439},"references/troubleshooting.md",{"path":547,"priority":439},"references/unsloth.md",{"path":549,"priority":500},"scripts/convert_to_gguf.py",{"path":551,"priority":500},"scripts/dataset_inspector.py",{"path":553,"priority":500},"scripts/estimate_cost.py",{"path":555,"priority":500},"scripts/hf_benchmarks.py",{"path":557,"priority":500},"scripts/train_dpo_example.py",{"path":559,"priority":500},"scripts/train_grpo_example.py",{"path":561,"priority":500},"scripts/train_sft_example.py",{"path":563,"priority":500},"scripts/unsloth_sft_example.py",{"basePath":256,"description":10,"displayName":258,"installMethods":565,"rationale":566,"selectedPaths":567,"source":317,"sourceLanguage":18,"type":259},{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-local-models/SKILL.md",[568,569,571,573],{"path":327,"priority":311},{"path":570,"priority":439},"references/hardware.md",{"path":572,"priority":439},"references/hub-discovery.md",{"path":574,"priority":439},"references/quantization.md",{"basePath":334,"description":335,"displayName":336,"installMethods":576,"rationale":577,"selectedPaths":578,"source":317,"sourceLanguage":18,"type":259},{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-paper-publisher/SKILL.md",[579,580,582,584,586,588,590,592],{"path":327,"priority":311},{"path":581,"priority":500},"examples/example_usage.md",{"path":583,"priority":439},"references/quick_reference.md",{"path":585,"priority":500},"scripts/paper_manager.py",{"path":587,"priority":500},"templates/arxiv.md",{"path":589,"priority":500},"templates/ml-report.md",{"path":591,"priority":500},"templates/modern.md",{"path":593,"priority":500},"templates/standard.md",{"basePath":342,"description":595,"displayName":344,"installMethods":596,"rationale":597,"selectedPaths":598,"source":317,"sourceLanguage":18,"type":259},"Look up and read Hugging Face paper pages in markdown, and use the papers API for structured metadata such as authors, linked models/datasets/spaces, Github repo and project page. Use when the user shares a Hugging Face paper page URL, an arXiv URL or ID, or asks to summarize, explain, or analyze an AI research paper.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-papers/SKILL.md",[599],{"path":327,"priority":311},{"basePath":390,"description":601,"displayName":392,"installMethods":602,"rationale":603,"selectedPaths":604,"source":317,"sourceLanguage":18,"type":259},"Use this skill when the user wants to build tool/scripts or achieve a task where using data from the Hugging Face API would help. This is especially useful when chaining or combining API calls or the task will be repeated/automated. This Skill creates a reusable script to fetch, enrich or process data.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-tool-builder/SKILL.md",[605,606,608,610,612,614,616,618],{"path":327,"priority":311},{"path":607,"priority":439},"references/baseline_hf_api.py",{"path":609,"priority":439},"references/baseline_hf_api.sh",{"path":611,"priority":439},"references/baseline_hf_api.tsx",{"path":613,"priority":439},"references/find_models_by_paper.sh",{"path":615,"priority":439},"references/hf_enrich_models.sh",{"path":617,"priority":439},"references/hf_model_card_frontmatter.sh",{"path":619,"priority":439},"references/hf_model_papers_auth.sh",{"basePath":374,"description":621,"displayName":376,"installMethods":622,"rationale":623,"selectedPaths":624,"source":317,"sourceLanguage":18,"type":259},"Track and visualize ML training experiments with Trackio. Use when logging metrics during training (Python API), firing alerts for training diagnostics, or retrieving/analyzing logged metrics (CLI). Supports real-time dashboard visualization, alerts with webhooks, HF Space syncing, and JSON output for automation.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-trackio/SKILL.md",[625,626,628,630],{"path":327,"priority":311},{"path":627,"priority":439},"references/alerts.md",{"path":629,"priority":439},"references/logging_metrics.md",{"path":631,"priority":439},"references/retrieving_metrics.md",{"basePath":414,"description":633,"displayName":416,"installMethods":634,"rationale":635,"selectedPaths":636,"source":317,"sourceLanguage":18,"type":259},"Trains and fine-tunes vision models for object detection (D-FINE, RT-DETR v2, DETR, YOLOS), image classification (timm models — MobileNetV3, MobileViT, ResNet, ViT/DINOv3 — plus any Transformers classifier), and SAM/SAM2 segmentation using Hugging Face Transformers on Hugging Face Jobs cloud GPUs. Covers COCO-format dataset preparation, Albumentations augmentation, mAP/mAR evaluation, accuracy metrics, SAM segmentation with bbox/point prompts, DiceCE loss, hardware selection, cost estimation, Trackio monitoring, and Hub persistence. Use when users mention training object detection, image classification, SAM, SAM2, segmentation, image matting, DETR, D-FINE, RT-DETR, ViT, timm, MobileNet, ResNet, bounding box models, or fine-tuning vision models on Hugging Face Jobs.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-vision-trainer/SKILL.md",[637,638,640,641,643,645,646,648,649,650,652,654],{"path":327,"priority":311},{"path":639,"priority":439},"references/finetune_sam2_trainer.md",{"path":533,"priority":439},{"path":642,"priority":439},"references/image_classification_training_notebook.md",{"path":644,"priority":439},"references/object_detection_training_notebook.md",{"path":537,"priority":439},{"path":647,"priority":439},"references/timm_trainer.md",{"path":551,"priority":500},{"path":553,"priority":500},{"path":651,"priority":500},"scripts/image_classification_training.py",{"path":653,"priority":500},"scripts/object_detection_training.py",{"path":655,"priority":500},"scripts/sam_segmentation_training.py",{"basePath":422,"description":657,"displayName":424,"installMethods":658,"rationale":659,"selectedPaths":660,"source":317,"sourceLanguage":18,"type":259},"Train or fine-tune sentence-transformers models across `SentenceTransformer` (bi-encoder; dense or static embedding model; for retrieval, similarity, clustering, classification, paraphrase mining, dedup, multimodal), `CrossEncoder` (reranker; pair scoring for two-stage retrieval / pair classification), and `SparseEncoder` (SPLADE, sparse embedding model; for learned-sparse retrieval). Covers loss selection, hard-negative mining, evaluators, distillation, LoRA, Matryoshka, and Hugging Face Hub publishing. Use for any sentence-transformers training task.",{"claudeCode":12},"SKILL.md frontmatter at skills/train-sentence-transformers/SKILL.md",[661,662,664,666,668,670,672,673,675,677,679,681,683,685,687,688,690,692,694,696,698,700,702,704,706,708,710,712],{"path":327,"priority":311},{"path":663,"priority":439},"references/base_model_selection.md",{"path":665,"priority":439},"references/dataset_formats.md",{"path":667,"priority":439},"references/evaluators_cross_encoder.md",{"path":669,"priority":439},"references/evaluators_sentence_transformer.md",{"path":671,"priority":439},"references/evaluators_sparse_encoder.md",{"path":531,"priority":439},{"path":674,"priority":439},"references/hf_jobs_execution.md",{"path":676,"priority":439},"references/losses_cross_encoder.md",{"path":678,"priority":439},"references/losses_sentence_transformer.md",{"path":680,"priority":439},"references/losses_sparse_encoder.md",{"path":682,"priority":439},"references/model_architectures.md",{"path":684,"priority":439},"references/prompts_and_instructions.md",{"path":686,"priority":439},"references/training_args.md",{"path":545,"priority":439},{"path":689,"priority":500},"scripts/mine_hard_negatives.py",{"path":691,"priority":500},"scripts/train_cross_encoder_distillation_example.py",{"path":693,"priority":500},"scripts/train_cross_encoder_example.py",{"path":695,"priority":500},"scripts/train_cross_encoder_listwise_example.py",{"path":697,"priority":500},"scripts/train_sentence_transformer_distillation_example.py",{"path":699,"priority":500},"scripts/train_sentence_transformer_example.py",{"path":701,"priority":500},"scripts/train_sentence_transformer_make_multilingual_example.py",{"path":703,"priority":500},"scripts/train_sentence_transformer_matryoshka_example.py",{"path":705,"priority":500},"scripts/train_sentence_transformer_multi_dataset_example.py",{"path":707,"priority":500},"scripts/train_sentence_transformer_static_embedding_example.py",{"path":709,"priority":500},"scripts/train_sentence_transformer_with_lora_example.py",{"path":711,"priority":500},"scripts/train_sparse_encoder_distillation_example.py",{"path":713,"priority":500},"scripts/train_sparse_encoder_example.py",{"basePath":406,"description":715,"displayName":408,"installMethods":716,"rationale":717,"selectedPaths":718,"source":317,"sourceLanguage":18,"type":259},"Use Transformers.js to run state-of-the-art machine learning models directly in JavaScript/TypeScript. Supports NLP (text classification, translation, summarization), computer vision (image classification, object detection), audio (speech recognition, audio classification), and multimodal tasks. Works in browsers and server-side runtimes (Node.js, Bun, Deno) with WebGPU/WASM using pre-trained models from Hugging Face Hub.",{"claudeCode":12},"SKILL.md frontmatter at skills/transformers-js/SKILL.md",[719,720,722,724,726,728,730,732],{"path":327,"priority":311},{"path":721,"priority":439},"references/CACHE.md",{"path":723,"priority":439},"references/CONFIGURATION.md",{"path":725,"priority":439},"references/EXAMPLES.md",{"path":727,"priority":439},"references/MODEL_ARCHITECTURES.md",{"path":729,"priority":439},"references/MODEL_REGISTRY.md",{"path":731,"priority":439},"references/PIPELINE_OPTIONS.md",{"path":733,"priority":439},"references/TEXT_GENERATION.md",{"sources":735},[736],"manual",{"closedIssues90d":244,"description":738,"forks":245,"homepage":739,"license":250,"openIssues90d":246,"pushedAt":247,"readmeSize":242,"stars":248,"topics":740},"Give your agents the power of the Hugging Face ecosystem","https://huggingface.co",[],{"classifiedAt":742,"discoverAt":743,"extractAt":744,"githubAt":744,"updatedAt":742},1778690772996,1778689536128,1778690770714,[221,222,220,217,218,219],{"evaluatedAt":254,"extractAt":293,"updatedAt":254},[],[749,778,806,828,852,870],{"_creationTime":750,"_id":751,"community":752,"display":753,"identity":759,"providers":763,"relations":771,"tags":774,"workflow":775},1778695116697.1858,"k17cfm34t6mxkmnmp1248gbkbs86nfm3",{"reviewCount":8},{"description":754,"installMethods":755,"name":757,"sourceUrl":758},"GGUF format and llama.cpp quantization for efficient CPU/GPU inference. Use when deploying models on consumer hardware, Apple Silicon, or when needing flexible quantization from 2-8 bit without GPU requirements.",{"claudeCode":756},"Orchestra-Research/AI-Research-SKILLs","GGUF Quantization","https://github.com/Orchestra-Research/AI-Research-SKILLs",{"basePath":760,"githubOwner":761,"githubRepo":762,"locale":18,"slug":221,"type":259},"10-optimization/gguf","Orchestra-Research","AI-Research-SKILLs",{"evaluate":764,"extract":769},{"promptVersionExtension":210,"promptVersionScoring":211,"score":275,"tags":765,"targetMarket":223,"tier":224},[221,766,220,767,768],"quantization","cpu-inference","model-optimization",{"commitSha":284,"license":770},"MIT",{"parentExtensionId":772,"repoId":773},"k17155ws9qc0hw7a568bg79sfd86max8","kd70hj1y80mhra5xm5g188j5n586mg18",[767,221,220,768,766],{"evaluatedAt":776,"extractAt":777,"updatedAt":776},1778696516392,1778695116697,{"_creationTime":779,"_id":780,"community":781,"display":782,"identity":788,"providers":793,"relations":800,"tags":802,"workflow":803},1778691799740.4775,"k17d0yq6vmmtzk249wz61kpa8n86mqrf",{"reviewCount":8},{"description":783,"installMethods":784,"name":786,"sourceUrl":787},"Use when the user is doing AI/ML work in a scientific domain — biology, chemistry, physics, astronomy, climate, genomics, materials science, medicine, ecology, energy, conservation, engineering, mathematics, scientific reasoning, drug discovery, protein design, weather modeling, theorem proving, single-cell, PDE solving, or anything similar. Hugging Science (huggingscience.co) is a curated catalog of scientific datasets, models, blog posts, and interactive Spaces; the `hugging-science` org on Hugging Face hosts community datasets, models, and demo Spaces. This skill helps you discover the right resource AND actually use it — loading datasets via `datasets`, running models via `transformers` or the HF Inference API, calling Spaces like BoltzGen via `gradio_client`, and citing blog posts for methodology. Trigger this skill whenever a user mentions a scientific ML task, asks for \"a dataset/model for X\" where X is a scientific topic, wants to fine-tune on scientific data, asks about protein / molecule / genome / climate / materials / astronomy / pathology / weather ML, or needs AI tools for research — even if they never say \"Hugging Science\" explicitly. The catalog is purpose-built for LLM agents (it ships an `llms-full.txt`); prefer it over generic web search for these tasks.",{"claudeCode":785},"K-Dense-AI/claude-scientific-skills","Hugging Science","https://github.com/K-Dense-AI/claude-scientific-skills",{"basePath":789,"githubOwner":790,"githubRepo":791,"locale":18,"slug":792,"type":259},"scientific-skills/hugging-science","K-Dense-AI","claude-scientific-skills","hugging-science",{"evaluate":794,"extract":799},{"promptVersionExtension":210,"promptVersionScoring":211,"score":275,"tags":795,"targetMarket":223,"tier":224},[222,796,279,219,797,278,798],"science","research","discovery",{"commitSha":284,"license":770},{"repoId":801},"kd79rphh5gexy91xmpxc05h5mh86mm9r",[279,798,222,278,219,797,796],{"evaluatedAt":804,"extractAt":805,"updatedAt":804},1778692838166,1778691799740,{"_creationTime":807,"_id":808,"community":809,"display":810,"identity":813,"providers":815,"relations":824,"tags":825,"workflow":826},1778695116697.1875,"k17csgsgze3c135qze92egp8gx86ma52",{"reviewCount":8},{"description":811,"installMethods":812,"name":220,"sourceUrl":758},"Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.",{"claudeCode":756},{"basePath":814,"githubOwner":761,"githubRepo":762,"locale":18,"slug":220,"type":259},"12-inference-serving/llama-cpp",{"evaluate":816,"extract":823},{"promptVersionExtension":210,"promptVersionScoring":211,"score":214,"tags":817,"targetMarket":223,"tier":224},[818,220,767,819,820,221,766,821,822],"inference-serving","apple-silicon","edge-deployment","non-nvidia","gpu-inference",{"commitSha":284},{"parentExtensionId":772,"repoId":773},[819,767,820,221,822,818,220,821,766],{"evaluatedAt":827,"extractAt":777,"updatedAt":827},1778696642747,{"_creationTime":829,"_id":830,"community":831,"display":832,"identity":836,"providers":841,"relations":846,"tags":848,"workflow":849},1778685991755.723,"k17evrcg4wz3mdbs7nab3vdxhn86nh0h",{"reviewCount":8},{"description":754,"installMethods":833,"name":757,"sourceUrl":835},{"claudeCode":834},"davila7/claude-code-templates","https://github.com/davila7/claude-code-templates",{"basePath":837,"githubOwner":838,"githubRepo":839,"locale":18,"slug":840,"type":259},"cli-tool/components/skills/ai-research/optimization-gguf","davila7","claude-code-templates","optimization-gguf",{"evaluate":842,"extract":845},{"promptVersionExtension":210,"promptVersionScoring":211,"score":214,"tags":843,"targetMarket":223,"tier":844},[221,766,220,767,768,819],"community",{"commitSha":284,"license":770},{"repoId":847},"kd71fzn4s7r0269fkw47wt670n86ndz0",[819,767,221,220,768,766],{"evaluatedAt":850,"extractAt":851,"updatedAt":850},1778688255792,1778685991755,{"_creationTime":853,"_id":854,"community":855,"display":856,"identity":858,"providers":861,"relations":866,"tags":867,"workflow":868},1778685991755.713,"k1733vggg1gqrsejz4938b9eq586ndp0",{"reviewCount":8},{"description":811,"installMethods":857,"name":220,"sourceUrl":835},{"claudeCode":834},{"basePath":859,"githubOwner":838,"githubRepo":839,"locale":18,"slug":860,"type":259},"cli-tool/components/skills/ai-research/inference-serving-llama-cpp","inference-serving-llama-cpp",{"evaluate":862,"extract":865},{"promptVersionExtension":210,"promptVersionScoring":211,"score":863,"tags":864,"targetMarket":223,"tier":844},85,[818,220,767,819,820,221,766],{"commitSha":284},{"repoId":847},[819,767,820,221,818,220,766],{"evaluatedAt":869,"extractAt":851,"updatedAt":869},1778687302086,{"_creationTime":871,"_id":872,"community":873,"display":874,"identity":876,"providers":877,"relations":885,"tags":886,"workflow":887},1778690773482.4866,"k17a3mmgvm5hj49twj487hp64186n2qa",{"reviewCount":8},{"description":481,"installMethods":875,"name":368,"sourceUrl":14},{"claudeCode":12},{"basePath":366,"githubOwner":222,"githubRepo":257,"locale":18,"slug":368,"type":259},{"evaluate":878,"extract":884},{"promptVersionExtension":210,"promptVersionScoring":211,"score":879,"tags":880,"targetMarket":223,"tier":224},100,[281,222,881,882,883],"mlops","data-management","model-management",{"commitSha":284},{"parentExtensionId":262,"repoId":289},[281,882,222,881,883],{"evaluatedAt":888,"extractAt":293,"updatedAt":888},1778691223210]