---
inference: false
---

# Model Card

Bunny is a family of lightweight multimodal models.

Bunny-pretrain-phi-2-eva is the pretrained weights for [bunny-phi-2-eva-lora](https://huggingface.co/BoyaWu10/bunny-phi-2-eva-lora), which leverages Phi-2 as the language model backbone and EVA-CLIP as the vision encoder.
It is pretrained on LAION-2M.

More details about this model can be found in [GitHub](https://github.com/BAAI-DCAI/Bunny).