# skills | Hugging Face | Knowledge Base | Instagit

GitHub Stars: 8.5k

Repository: https://github.com/huggingface/skills

---

## Articles

### [Integrating Skills Plugin with Claude Code, OpenAI Codex, Gemini CLI, and Cursor](/huggingface/skills/integrating-skills-plugin-claude-code-openai-codex-gemini-cli-cursor)

Integrate the Hugging Face Skills plugin with Claude Code, OpenAI Codex, Gemini CLI, and Cursor. Discover and execute AI ML skills using manifest files for enhanced agent capabilities.

- Tags: how-to-guide
- Published: 2026-03-08

### [Handling Authentication for Pushing Models and Datasets to Hugging Face Hub](/huggingface/skills/handling-authentication-pushing-models-datasets-hugging-face-hub)

Learn to handle Hugging Face Hub authentication for pushing models and datasets. Securely log in using tokens and environment variables for seamless uploads.

- Tags: how-to-guide
- Published: 2026-03-08

### [Extracting Evaluation Tables from READMEs to YAML Format in Hugging Face Skills](/huggingface/skills/extracting-evaluation-tables-readmes-yaml-format)

Convert evaluation tables from Hugging Face model READMEs to YAML format. The hugging-face-evaluation skill uses token matching and auto detection for standardized model-index YAML.

- Tags: how-to-guide
- Published: 2026-03-08

### [Configuring PEP 723 Inline Dependencies in UV Scripts: The Complete Guide](/huggingface/skills/configuring-pep-723-inline-dependencies-uv-scripts)

Learn to configure PEP 723 inline dependencies in UV scripts for automatic package installation. Master dependency management in your Python projects with this complete guide.

- Tags: how-to-guide
- Published: 2026-03-08

### [Checking for Existing PRs Before Creating New Evaluation PRs in Hugging Face Skills](/huggingface/skills/checking-existing-prs-before-creating-evaluation-prs)

Avoid duplicate work by checking for existing PRs before creating new evaluation PRs in Hugging Face Skills. Use the get_open_prs function to query the API.

- Tags: how-to-guide
- Published: 2026-03-08

### [Running Evaluation Jobs with inspect-ai on Hugging Face Using UV Scripts](/huggingface/skills/running-evaluation-jobs-inspect-ai-hugging-face-uv-scripts)

Easily run inspect-ai evaluation jobs on Hugging Face. Our UV scripts support vLLM and Transformers backends, making evaluations simple and secure.

- Tags: how-to-guide
- Published: 2026-03-08

### [Using Unsloth for Faster Training with Reduced VRAM: A Complete Guide](/huggingface/skills/using-unsloth-faster-training-reduced-vram)

Accelerate your AI training with Unsloth. Cut VRAM usage by 60% and double speeds using optimized kernels and 4-bit quantization. Learn how in this complete guide.

- Tags: how-to-guide
- Published: 2026-03-08

### [Supported Template Formats for Dataset Creation in Hugging Face Skills](/huggingface/skills/supported-template-formats-dataset-creation)

Discover supported template formats for Hugging Face dataset creation. Learn about chat, classification, qa, completion, tabular, and custom formats for efficient dataset building.

- Tags: tutorial
- Published: 2026-03-08

### [Creating and Managing Hugging Face Hub Datasets with System Prompts](/huggingface/skills/creating-managing-hugging-face-hub-datasets-system-prompts)

Easily create and manage Hugging Face Hub datasets with system prompts. Configure and populate datasets directly in repository config files using JSON templates.

- Tags: how-to-guide
- Published: 2026-03-08

### [Estimating Cost and Runtime of Hugging Face Training Jobs Before Submission](/huggingface/skills/estimating-cost-runtime-training-jobs-submission)

Estimate Hugging Face training job costs and runtime before submission. Learn how the estimate_cost.py script predicts expenses and duration using model size, hardware pricing, and empirical formulas.

- Tags: performance
- Published: 2026-03-08

### [Setting Appropriate Timeout Values for Hugging Face Training Jobs](/huggingface/skills/setting-appropriate-timeout-values-training-jobs)

Learn how to set effective timeout values for your Hugging Face training jobs. Extend your runtime by 30% to ensure checkpoint saving and prevent premature termination.

- Tags: best-practices
- Published: 2026-03-08

### [Common Failure Modes for Hugging Face Training Jobs: A Complete Troubleshooting Guide](/huggingface/skills/common-failure-modes-hugging-face-training-jobs-troubleshooting)

Troubleshoot common Hugging Face training job failures like OOM errors and dataset issues. Learn solutions for gradient checkpointing, streaming, and hub limits.

- Tags: how-to-guide
- Published: 2026-03-08

### [Dataset Column Mapping for DPO Training: Handling Non-Standard Column Names in Hugging Face Skills](/huggingface/skills/handling-dataset-column-mapping-dpo-training-non-standard-names)

Effortlessly map dataset columns for DPO training with Hugging Face Skills. Use dataset_inspector.py to auto-detect and remap non-standard names to prompt chosen and rejected formats.

- Tags: how-to-guide
- Published: 2026-03-08

### [Importing Benchmark Scores from Artificial Analysis API to Hugging Face Model Cards](/huggingface/skills/importing-benchmark-scores-artificial-analysis-api-model-cards)

Automate importing benchmark scores from Artificial Analysis API to Hugging Face model cards. Keep leaderboards updated and showcase your model performance effortlessly with the hugging-face-evaluation skill.

- Tags: how-to-guide
- Published: 2026-03-08

### [Difference Between Inspect-AI and Lighteval for Model Evaluation](/huggingface/skills/difference-inspect-ai-lighteval-model-evaluation)

Compare Inspect-AI and Lighteval for model evaluation. Discover Inspect-AI's safety benchmarks and Lighteval's high-throughput multi-tasking for efficient model assessment.

- Tags: deep-dive
- Published: 2026-03-08

### [Running Custom Model Evaluations with vLLM and LightEval on Hugging Face Jobs](/huggingface/skills/running-custom-model-evaluations-vllm-lighteval-hugging-face-jobs)

Effortlessly run custom model evaluations with vLLM and LightEval on Hugging Face Jobs. Auto-install dependencies and provision GPUs using the hf jobs uv run command.

- Tags: tutorial
- Published: 2026-03-08

### [Adding Evaluation Results to Model Cards Using Model-Index YAML: A Complete Guide](/huggingface/skills/adding-evaluation-results-model-cards-model-index-yaml)

Easily add evaluation results to Hugging Face model cards with model-index YAML. Automate benchmark scores from READMEs or API data using the evaluation manager CLI.

- Tags: how-to-guide
- Published: 2026-03-08

### [Querying and Transforming Hugging Face Datasets with DuckDB and hf://: A Complete Guide](/huggingface/skills/querying-transforming-hugging-face-datasets-duckdb-hf-protocol)

Query Hugging Face datasets with DuckDB using the hf:// protocol. Transform and export data seamlessly without downloading files. A complete guide.

- Tags: how-to-guide
- Published: 2026-03-08

### [Using Trackio for Real-Time Training Metrics on Hugging Face Spaces: A Complete Guide](/huggingface/skills/using-trackio-real-time-training-metrics-hugging-face-spaces)

Learn how to use Trackio for real-time training metrics on Hugging Face Spaces. Monitor loss, learning rate, and system data live with this comprehensive guide.

- Tags: tutorial
- Published: 2026-03-08

### [How to Convert Fine-Tuned Models to GGUF for Ollama and llama.cpp](/huggingface/skills/how-to-convert-fine-tuned-models-gguf-ollama-llamacpp)

Convert fine-tuned models to GGUF for Ollama and llama.cpp. Merge LoRA, build llama.cpp tools, and run the script to generate quantized GGUF files for efficient local AI deployment.

- Tags: how-to-guide
- Published: 2026-03-08

### [Recommended Hardware Configurations for Training Different Model Sizes on Hugging Face Jobs: The Complete Guide](/huggingface/skills/recommended-hardware-configurations-training-model-sizes-hugging-face-jobs)

Find optimal hardware for Hugging Face Jobs using our guide. Choose GPU flavors based on model size—from T4 for <1B models to A100 for >13B—to maximize cost efficiency and memory.

- Tags: tutorial
- Published: 2026-03-08

### [How to Validate Dataset Format for GPU Training Jobs with Hugging Face Skills](/huggingface/skills/how-to-validate-dataset-format-gpu-training-jobs)

Validate dataset format for GPU training jobs with Hugging Face Skills. Use Dataset Manager for schema checks and Dataset Inspector for remote compatibility with SFT DPO GRPO and KTO.

- Tags: how-to-guide
- Published: 2026-03-08

### [Best Practices for Saving Trained Models to Hugging Face Hub in Jobs](/huggingface/skills/best-practices-saving-trained-models-hugging-face-hub-jobs)

Learn best practices for saving trained models to Hugging Face Hub in jobs. Discover how to use HF_TOKEN for persistent artifact storage in ephemeral job containers.

- Tags: best-practices
- Published: 2026-03-08

### [How to Configure TRL Training Jobs on Hugging Face Jobs: A Complete Guide to SFT, DPO, and GRPO](/huggingface/skills/how-to-configure-trl-training-jobs-hugging-face-jobs-sft-dpo-grpo)

Configure TRL training jobs on Hugging Face Jobs using hf_jobs MCP helper. Run SFT, DPO, and GRPO fine-tuning with `a10g-large` GPUs and Hub authentication for efficient distributed training.

- Tags: how-to-guide
- Published: 2026-03-08