DavidUdell / sparse_circuit_discovery

Circuit discovery in GPT-2 small, using sparse autoencoding
MIT License
7 stars 1 forks source link

Add a THINNING_FACTOR YAML constant for full-scale graphing. #25

Closed DavidUdell closed 10 months ago

DavidUdell commented 10 months ago

Full-scale graphs are so large as to be unreadable, whether or not their interp data is good. So it's better to have a readable subset of them, chosen at random, than to have the whole thing so densely cluttered that it cannot be read.

This would be a YAML constant that filters the ablation dims list down to a subset. It would happen before DIMS_GRAPHED_LIST is invoked, so that the two are mutually exclusive.