7 ms·
It probably gives them confidence that they can accurately see a thing even though they don't know what that thing is. I could also imagine a lot of safety aro
by jayd16 2y ago
It probably gives them confidence that they can accurately see a thing even though they don't know what that thing is.
I could also imagine a lot of safety around leaving things outside of the current task alone so you might have to bend over backwards to get new objects worked on.
- thomastjeffery 2y agoThere is no such thing as "thing" here. These models are trained such that the given conditions (the visual input and the text prompt) will be continued with a desirable continuation (motor function over time). The only dimension accuracy can apply to is desirability.
- jayd16 2y agoYou don't think there's any segmentation going on?
- thomastjeffery 2y agoImplicitly, maybe. Does that matter if you don't know where?