We're using it!
#1
by
jaredquinn
- opened
Thank you for your pre-trained model! We're using it to caption the SSTV images for a mastodon bot at https://botsin.space/@aregsstv
You may notice a lot of noise images come through accidently there; we can help provide some data to recognise when noise is noise and not "a vase of colorful colorful colorful colorful colorful..."
Hi Jared, Thanks for your comment. I've noticed what you've already noticed. Actually, there are some noises in the images. For example, when I upload a picture of a man wearing caps, our image caption model starts repeating some of the captions. We need enough data to fix this. If you can help me by providing data, that would be great.