Instruction-finetuning Hyperparameters

zzxslp / SoM-LLaVA

[COLM-2024] List Items One by One: A New Data Source and Learning Paradigm for Multimodal LLMs

115 stars 2 forks source link

Closed xirui-li closed 4 hours ago

xirui-li commented 4 hours ago

Thanks for sharing your awesome dataset!

I am wondering whether you freeze the vision tower when you finetune the pretrained Llava. Any useful information would be greatly appreciated!

xirui-li commented 4 hours ago

Figured it out in Llava project . Thanks!