huggingface/transformers
Model-definition framework for text, vision, audio, and multimodal ML
- Chart
- Not on a chart right now
- Stars
- 169.2k
What it is
Transformers is a model-definition framework for inference and training across text, computer vision, audio, video, and multimodal machine learning models. It centralizes model definitions so they work across major training frameworks and inference engines.
Who it is for
Researchers, engineers, and developers building or running pretrained machine learning models will find it useful.
Why it is trending
It released v5.23.0 on November 13, 2026, and gained 262 stars today, bringing its total to 169,208 stars. It stands out as a shared model-definition layer used by tools such as vLLM, SGLang, Axolotl, and Unsloth.
Main features
- High-level Pipeline API for text, audio, vision, and multimodal inference
- Access to over 1M pretrained model checkpoints on the Hugging Face Hub
- Compatibility across PyTorch, JAX, TF2.0, and major training or inference frameworks
- Command-line chat interface for running models directly
Summarized on November 18, 2026 from the README and repository data. It can be wrong; check the source before relying on it. Report a wrong summary