search

microsoft / LLaVA-Med

Large Language-and-Vision Assistant for Biomedicine, built towards multimodal GPT-4 level capabilities.
Other
1.29k stars 148 forks source link

Candidate file? #26

Open hellocym opened 7 months ago

hellocym commented 7 months ago

I found that run_eval.py needs a cadidate file to evaluate, but I couldn't find a way to generate the file. Is this file necessary? And how can I generate the file?

M3Dade commented 7 months ago

@hellocym I also found the problem. Have you solved it?

hellocym commented 6 months ago

@hellocym I also found the problem. Have you solved it?

Unfortunately, no :( By the way, I managed to create my own evaluation script on closed set, but I failed to reproduce the accuracy reported in the paper.

LiangXin1001 commented 4 months ago

This is my modified version of run_eval.py based on the source code. I hope everyone can check if there are any issues with the code.

M3Dade commented 4 months ago

这是我根据源代码修改的run_eval.py版本。希望大家能检查一下代码是否有问题。 @LiangXin1001 It even can't work...

zhouqiji commented 2 months ago

Same issue...