Surprisingly solid.

#1
by Todokete - opened

I stumbled across this while searching for "captioning" in spaces, and I gotta say, it's pretty good for the size.

What dataset(s) did you use to train this?

Hi, Im using a synthetic dataset with llama 4 Maverick
https://huggingface.co/datasets/Andres77872/findit_v0.2_safe

The performance on quote unquote "unsafe" images is quite impressive too. Am I right to assume the publicly available dataset is only part of what was trained on?

yes, is just the safe subset, idk if hugging face allow nsfw, 200k sfw and 200k nsfw
i want a model for all xD, tagger, caption and vector embedding for reverse image search

Andres77872 changed discussion status to closed

Sign up or log in to comment