question on python env setup for evaluation

[Tsingularity]

Hi, thanks for open-soucing the great work!

I'm a bit lost when setting up the code for evalation, because I saw there are two python env setup in Readme:

1. https://github.com/Karine-Huang/T2I-CompBench?tab=readme-ov-file#installing-the-dependencies
2. https://github.com/Karine-Huang/T2I-CompBench?tab=readme-ov-file#evaluation

Just wondering which environment should I use for evaluating T2I models? I tried using share4v but it seems to give me endless python library errors.

Thanks!

[Karine-Huang]

Hi!
If you want to use MiniGPT-4 or ShareGPT-4V, you can use the second link based on their repositories (https://github.com/Karine-Huang/T2I-CompBench?tab=readme-ov-file#evaluation). Otherwise, the first link is enough.