llziss4ai

There is a [repo](https://github.com/PKU-YuanGroup/LanguageBind) that uses LoRA to bind modalities to language.

We can use absolute depth in meters to inference by this [repo](https://github.com/PKU-YuanGroup/LanguageBind)

**It seems more efficient to align the text directly.** 

> Hi authors, > > Thanks for the great ImageBind! I noticed that for LLVIP and Ego4d-IMU, you constructed the train/val split by yourself. I wonder if you will release...