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.

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.

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.

Nowadays, you can get a free VM from some providers (GCP) and there are lots of sites where you can deploy a static website for free (github pages). I'd argue deploying something for free is easier than ever.