Montgomery says that Reddit will use the new Florence-powered APIs to generate captions for images on its platform, creating “alt text” so users with vision challenges can better follow along ...

The photo above was generated by DALL-E from the text prompt “an illustration of a baby daikon radish in a tutu walking a dog.” DALL-E uses a 12-billion parameter version of GPT-3, and like ...