We trained a crosscoder of width 16,384 on the residual stream activations from the middle layer of the Gemma-2 2B base and IT models.
I don't understand the training process here, as well as the mini-paper from Anthropic. How do you train one crosscoder on the residual stream from two different models?
I don't understand the training process here, as well as the mini-paper from Anthropic. How do you train one crosscoder on the residual stream from two different models?