And openAI has the full one, 10x stronger. This one I tested only has 300 million parameters and trained on only 67 million text-image pairs. The full GLIDE AI has 250M pairs trained on and 3.5 billion parameters. And humans aren't filtered out in that version. It also uses the same compute as DALL-E but less RAM since it is 3.5B parameters! And, is preferred over DALL-E. That's pretty big stuff, I guess. I'm not sure if it is better than DALL-E for text 2 image, though I've not tried image completion with DALL-E either, so I cannot compare them fully.
------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Te6817de606bbe208-M52d7b2df8688385b6aa097a4 Delivery options: https://agi.topicbox.com/groups/agi/subscription
