Jaehyun Jun's picture

Jaehyun Jun

btjhjeon

·

https://btjhjeon.github.io/

btjhjeon

AI & ML interests

Multimodal

Recent Activity

upvoted a paper 3 days ago

FastVLM: Efficient Vision Encoding for Vision Language Models

upvoted a paper 3 days ago

Descriptive Caption Enhancement with Visual Specialists for Multimodal Perception

updated a collection 3 days ago

View all activity

Organizations

btjhjeon's activity

New activity in Qwen/Qwen2-VL-7B-Instruct 3 months ago

when i use torch.float16，i face this problem probability tensor contains either `inf`, `nan` or element < 0

#20 opened 4 months ago by

New activity in google/siglip-so400m-patch14-384 4 months ago

Why is the image size designed to be 384, whereas the patch size is designed to be 14, when 384 is not divisible by 14?

#4 opened 9 months ago by