apple/ml-aim
Apple's AIM is a family of large-scale autoregressive vision encoders pre-trained with a multimodal self-supervised objective.
Not currently ranked — collecting fresh signals.
star history
This repository provides code and pre-trained model checkpoints for AIMv1 and AIMv2, two large vision encoder models trained using autoregressive pre-training at scale. AIMv2 is multimodal and outperforms OAI CLIP and SigLIP on benchmarks. The project supports multiple frameworks (PyTorch, JAX, MLX) and offers pre-trained backbones for the research community.
Frequently asked
- What is apple/ml-aim?
- Apple's AIM is a family of large-scale autoregressive vision encoders pre-trained with a multimodal self-supervised objective.
- Is ml-aim open source?
- Yes — apple/ml-aim is an open-source project tracked on heatdrop.
- What language is ml-aim written in?
- apple/ml-aim is primarily written in Python.
- How popular is ml-aim?
- apple/ml-aim has 1.4k stars on GitHub.
- Where can I find ml-aim?
- apple/ml-aim is on GitHub at https://github.com/apple/ml-aim.