LLaVA-VL / LLaVA-Interactive-Demo

LLaVA-Interactive-Demo
https://llava-vl.github.io/llava-interactive/
Apache License 2.0
348 stars 25 forks source link

Using LLaVA Plus?/ #7

Closed shersoni610 closed 7 months ago

shersoni610 commented 10 months ago

Hello,

Does this code use LLaVAplus ?

thaks

ChunyuanLI commented 10 months ago

No. LLaVA-Interactive and LLaVA-Plus are independent at this moment.

LLaVA-Interactive employs LLaVA-1.5 for visual chat. It is a demo that combines three existing vision-language models without anything training.

In LLaVA-Interactive, the tools/skills are manually selected by the humans, and tool execution results are directly presented to humans; In LLaVA-Plus, the tools/skills are automatically selected by LMM and the tool execution results are further aggregated by the LMM to present to humans.

mattmazzola commented 7 months ago

Given there has not been any more activity on this issue and Chunyuan's response from many months ago, I assume the question has been answered sufficiently and the issue can be closed.