alt Hacker NewsI am curious what's the actual formula.
I mean, there so many headers and layers, it is tricky to make a choice that will resonate with our intuition . Is it some weighted average? Or maybe ablation test?
Replies
visarga • today at 7:18 PM
If you want quick access look at google images for "transformer attention formula" there are some interesting depictions
It's really simple, basically just the magnitude of the value vector, weighted by QK dot product, summed across all attention heads and layers.
When I started, I expected I'd have to experiment a lot to find something comprehensible. But this simple computation can already show some patterns.