CircleRadon / TokenPacker

The code for "TokenPacker: Efficient Visual Projector for Multimodal LLM".
148 stars 6 forks source link

代码疑问 #5

Closed necrophagists closed 1 month ago

necrophagists commented 2 months ago

image 为什么这里要交换2 3 维度呢?

LiWentomng commented 2 months ago

@necrophagists 这里交换是为了保持kernel size的顺序与h w一致。

Eric-is-good commented 2 weeks ago

@necrophagists 请问你代码跑起来了吗,我是指推理部分,cli.py,我有一些问题想请教