facebookresearch/perception_models
Meta AI releases Perception Encoder and Perception Language Model for multimodal image, video, and audio understanding.
Not currently ranked — collecting fresh signals.
star history
This repository hosts Perception Encoder (PE) for encoding images, video, and audio into embeddings, and Perception Language Model (PLM) for multimodal decoding and generation. The models achieve state-of-the-art results on perception benchmarks and are integrated into popular libraries including Hugging Face transformers and timm. Released checkpoints include multiple model sizes and variants tuned for different use cases.
Frequently asked
- What is facebookresearch/perception_models?
- Meta AI releases Perception Encoder and Perception Language Model for multimodal image, video, and audio understanding.
- Is perception_models open source?
- Yes — facebookresearch/perception_models is open source, released under the Apache-2.0 license.
- What language is perception_models written in?
- facebookresearch/perception_models is primarily written in Jupyter Notebook.
- How popular is perception_models?
- facebookresearch/perception_models has 2.3k stars on GitHub.
- Where can I find perception_models?
- facebookresearch/perception_models is on GitHub at https://github.com/facebookresearch/perception_models.