@matt @simon yep, but that's a different topic. Here I was curious on what it takes in addition to the prompt for the model to know "coordinates". Eiter it is fed in as tokens or there is a parallel input.

@simon thanks, good and very insightful text. Minor typo in "may hae caught".
I am again intrigued by the way LLMs pre-prompt text works by explaining things politely and the LLM then has a new capability. But in addition they must have given coordinates during training on images, I would assume. What do you think?

@jon @athos @[email protected] what is the plan towards profitability? I confess not to have heard of @vivaldi yet. Sounds nice and a cool team! Company owned all is great. But you need money sooner or later. As an alternative to Firefox, I'll recommend from now on.