Publications

I play with vision-language models and cat pictures.

3 serious papers • many cat photos • countless failed experiments

2050

CatVLM: When Your Dataset Is 99% Cat Photos

CVPR'50🏆 Oral CatVLM: When Your Dataset Is 99% Cat Photos

Professor Meow*, Sleepy GPU

Paper / Code / Dataset / Blog (Zhihu) / YouTube / Bilibili

TL;DR

We train a VLM almost entirely on cat photos. Surprisingly, it still works on normal images.

Do Models Prefer Cats or Dogs? A Minimal Study

arXiv'2050 Do Models Prefer Cats or Dogs? A Minimal Study

Professor Meow, Doggo Reviewer 2

Paper / Code (coming soon)

TL;DR

We compare model preferences on simple cat vs dog prompts.

A Short Survey of Cat-Themed Benchmarks

Under Review A Short Survey of Cat-Themed Benchmarks

Professor Meow

Paper (under review)