NathanNeuro

Here's my function to look at the importances of individual neurons using your causal trace methodology, and then plot them. I don't know if you'll actually want to merge this...

prob contributions by individual neuron rather than layer

2

working on some code to expand your causal trace demo to look at the prob contributions by individual neuron rather than layer. It's pretty messy still but if you're interested...

Plot top k most-probable logits in logit lens

1

I'd like to see not just the single absolute probable token, but the top k most probable tokens.

bad example

1

I think presenting an example of jsonlines where you have a 'technically acceptable' example of a csv style jsonlines is a bad idea. The main advantage of jsonlines is that...

Personally, I'd like it if a dictionary like this included every word that was at least 7 letters long, differed from more common words by at least Levenshtein distance >=...