LVLM Implementation

Hi,

Thanks for your excellent work! I observed that the output of the model is an embedding. Is there an implementation of LVLM that allows the model to output actions in natural language?

Looking forward to reply.