Post

What do LLMs see? I wrote a lil' tool that extracts the attention matrices out of open models and creates this typing visual, with each token's opacity changing according to its average attention score as the prompt progresses. Dimmer words are considered less important to the model.

that’s very cool! Have you tried/considered using a colour scale, eg temperature colours (red, yellow, blue, white) so that subtle variations are easier to pick out?