利用 Meta 的 ImageBind 训练出来的多模态模型。
只使用了 文本-图像 数据进行微调就获得了很好的多模态效果。https://vxtwitter.com/yixuan_su/status/1661064018868551691

----------------------
vxTwitter
Yixuan Su (@yixuan_su)

We are super excited to share PandaGPT, the first foundation model capable of instruction-following data across six modalities, without the need of explicit supervision. [1/n]

Project Page: https://panda-gpt.github.io/…

----------------------

via AI News - Telegram Channel