This is great, I've read multiple books and watched videos about the attention mechanism. Now that I understand it, this is the clearest example I've seen on how attention works.
I don't know much about LLMs but does that mean you have N^2 computation with the context size since every token needs to track how it relates to every other token?
Yes, except no with the KV cache. Because tokens aren't modified by future tokens you can cache the meaning of previous tokens. This makes the total effort linear over the entire context (or constant per forward pass).
I don't disagree with that. I did add an entire caveat paragraph there.
To me, it's more of a neat visualization, not something that can be used to interpret LLM behavior. Even with a lot of simplification, it can show some interesting patterns.
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?
This is great, I've read multiple books and watched videos about the attention mechanism. Now that I understand it, this is the clearest example I've seen on how attention works.
neat, combining info from two phrases is hard to see without such a tool.
are you worried later-layer attention gets drowned out by earlier layers just because there are more of them contributing to the sum?
Hmm, I might try to add some controls to limit which layers get summed up. It might be able to reveal more patterns.
Right now only simple correlations are visible.
I like the visualisation. Pretty cool
I don't know much about LLMs but does that mean you have N^2 computation with the context size since every token needs to track how it relates to every other token?
Yes, except no with the KV cache. Because tokens aren't modified by future tokens you can cache the meaning of previous tokens. This makes the total effort linear over the entire context (or constant per forward pass).
I see and is there only 1 layer of relations?
Or does it accumulate the relations like A relates to B, so also add in B's relations
For full self attention yes
I highly question this simplistic idea of high vector magnitude = high influence.
You get what you pay for. If you want to think harder and get more, https://transformer-circuits.pub/2025/attention-qk/index.htm...
I don't disagree with that. I did add an entire caveat paragraph there.
To me, it's more of a neat visualization, not something that can be used to interpret LLM behavior. Even with a lot of simplification, it can show some interesting patterns.
Same I dont get it just, could you clarify it
I 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?
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.
Nice! Sometimes the simplest approaches work the best.
If you want quick access look at google images for "transformer attention formula" there are some interesting depictions