Développement
agents-course
Hugging Face's free course on AI agents — smolagents, LangGraph and LlamaIndex, with a final benchmark
★ 33,3 k étoilesApache-2.0
Voir le guide →datasets est un projet de développement open source, créé par huggingface (huggingface/datasets sur GitHub). Il compte 22 042 étoiles sur GitHub et 3 506 forks, il est publié sous la licence Apache-2.0 et a été mis à jour pour la dernière fois sur GitHub le 8 octobre 2026.
Hugging Face's library for loading and processing datasets for AI models in one line
Find a dataset on the Hugging Face Hub, or prepare local files (CSV, JSON, images, audio) to load.
Install the datasets package with pip or conda - no repository clone needed for normal use.
Add optional extras for audio, vision or framework integration (PyTorch, TensorFlow, JAX) as needed.
Call load_dataset, inspect the first example, and process it with map() or streaming.
Load Hub or local datasets with load_dataset and get a typed, memory-mapped Dataset object.
Stream datasets without downloading them, so you can start training on huge corpora immediately.
Apply map, filter and formatting operations backed by Arrow for speed and low memory use.
Handle text, audio, images, video, PDFs and even 3D medical (NIfTI) data with dedicated helpers.
Feed datasets directly into PyTorch, TensorFlow or JAX training loops with optional extras.
Build datasets from your own CSV, JSON, Parquet, text or folder structures.
Get the source code from GitHub and enter the project directory.
git clone https://github.com/huggingface/datasets.git
cd datasetsThe base install covers text and tabular data; extras add audio, vision and framework support.
pip install datasets
# optional: pip install datasets[audio]Use the Quick Start pattern from the README: load, then read the first training example.
from datasets import load_dataset
ds = load_dataset("rajpurkar/squad")
print(ds["train"][0]) # see README quick startIt is maintained by Hugging Face and a large open-source community on GitHub.
Yes - the library is Apache-2.0 licensed and free. Individual datasets on the Hub carry their own licenses, so check each dataset card.
No paid key is needed. A free Hugging Face account/token is only required for gated or private datasets.
Compared with loading data by hand with Pandas, Datasets adds Hub integration, Arrow-backed memory mapping, streaming and standard splits - the tradeoff is learning its API for simple one-off jobs.
Run pip install datasets, then call load_dataset with a Hub dataset name and print the first example, following the README Quick Start.
Hugging Face's free course on AI agents — smolagents, LangGraph and LlamaIndex, with a final benchmark
A command-line AI coding assistant powered by DeepSeek Coder models.
NVIDIA NeMo Data Designer: build validated, multimodal synthetic datasets from scratch or from seed data.
MCPJam Inspector: chat with, inspect, debug, and evaluate MCP servers across clients and models.
Fine-tune huge models on small hardware with LoRA and other efficient methods
A context graph for the AI-native SDLC — so agents and humans share one map of the codebase.
Les autres projets de la catégorie Développement avec le plus d'étoiles du catalogue LibreAI — la sélection réaliste d'alternatives si datasets ne vous convient pas.
The open source coding agent
The model-definition framework behind modern AI — thousands of pretrained transformers for text, vision, and audio.
Anthropic's agentic coding tool that lives in your terminal and works across your whole codebase
Cross-platform desktop all-in-one manager for Claude Code, Codex, OpenCode and other coding agents
OpenAI's lightweight coding agent that runs locally in your terminal
Projets de la catégorie Développement proches dans le catalogue, les plus proches en étoiles de datasets — continuez à explorer de part et d'autre.
A desktop GUI and toolkit for Claude Code - agents, sessions, analytics and MCP
Fine-tune huge models on small hardware with LoRA and other efficient methods
A 25MB database client for 100+ databases — with a built-in AI assistant, MCP server and CLI
Reverse engineer anything with agents - from app behavior down to native binaries
TOON: a compact, human-readable serialization of JSON that saves tokens in LLM prompts
Self-hosted TikTok and Douyin scraper API - no-watermark downloads, MCP and web console
Gratuit et ouvert : récupérez le code source, mettez une étoile au dépôt et commencez à créer.