Another multimodal models？

[hhy150]

Hello, may I ask if there are any other multimodal models that can be loaded besides this model?
An example is : luodian/OTTER-MPT1B-RPJama-Init
I saw the use of llama llm and other vision encoders in the paper, but I don't quite understand how to decouple the luodian/OTTER-MPT1B-RPJama-Init model with LLM and Vision Encoder? Is there any way to achieve it?

[Jazzcharles]

We only adopt various vision encoders in our retrieval model, while fixing the vision encoder (CLIP-Large) in the captioning model.

[hhy150]

ok, thank you for your reply.