Vision language models

Just by seeing the starting state and end goal, AI models could predict how to complete a task by filling in the unknown steps and focusing only on the actions that matter.

A consistent rulebook of prompts could be the key to advancing artificial intelligence models, enhancing their image classification performance even when dealing with unknown objects.