Same as tvp-base, but finetuned on ANet dataset.
The TVP model was pretrained on public datasets:
Create images in seconds. No sign-up, no paywall, no setup.