Search tools, frameworks, libraries, categories and domains
Run the same PyTorch training code on CPU, multi-GPU, TPU, FSDP or DeepSpeed with minimal changes.
pip install accelerate
Package accelerate, last 26 weeks.
accelerate
Other tools in the same categories.
huggingface/transformers
Hugging Face's reference library for defining, loading and fine-tuning pretrained text, vision and audio models.
huggingface/huggingface_hub
The main registry for open models, datasets and Spaces, plus the Python client for uploading and downloading.
huggingface/datasets
Load, stream and process large ML datasets from the Hub with Apache Arrow-backed memory mapping.
huggingface/diffusers
Hugging Face's library of pretrained diffusion pipelines for image, video and audio generation.
huggingface/peft
Parameter-efficient fine-tuning methods such as LoRA and QLoRA for adapting large models cheaply.
huggingface/trl
Hugging Face library for post-training LLMs with SFT, DPO, GRPO and other RL-based methods.
huggingface/tokenizers
Fast Rust-backed implementations of BPE, WordPiece and Unigram tokenizers for training and inference.
huggingface/sentence-transformers
Standard library for computing and fine-tuning text embeddings and rerankers for search and RAG.
huggingface/transformers.js
Run Hugging Face models (NLP, vision, audio) directly in the browser via ONNX Runtime and WebGPU.