v4.49.0-AyaVision
📦 transformersView on GitHub →
✨ 4 features🔧 3 symbols
Summary
This release introduces Aya Vision 8B and 32B, multilingual multimodal models combining SigLIP-2 vision encoders with Cohere language models, available via a specialized transformers release tag.
Migration Steps
- Install the specific release tag using: pip install git+https://github.com/huggingface/transformers@v4.49.0-AyaVision
✨ New Features
- Added support for Aya Vision 8B and 32B models, state-of-the-art multilingual multimodal models.
- Integration of Siglip2-so400-384-14 vision encoder with Cohere CommandR-7B and Aya Expanse 32B language models.
- Multimodal capabilities (image and text) across 23 languages.
- Support for Aya Vision chat templates via AutoProcessor.