LLaVA-VL / LLaVA-Plus-Codebase

LLaVA-Plus: Large Language and Vision Assistants that Plug and Learn to Use Skills
https://llava-vl.github.io/llava-plus/
Apache License 2.0
698 stars 53 forks source link

[Feature request] Any plans to release larger model's weight? #19

Open xiaosu-zhu opened 10 months ago

xiaosu-zhu commented 10 months ago

feature

Thanks for your good job. I have tried your current code and model weight and successfully run on the server (well, some bugs are found for frontend interactions). Most times the model could correctly parse inputs and utilizes correct tool. But sometimes it may only produce meaningless texts. This may be the issue of model capacity. So I wonder if it is possible to publish the larger model's weight for comparison?