Where can one read in more details about your toy experiment setup and the large scale one? Like how did you define the causality score and why it has been concluded that causal confusion reduces by depth and more data? Do you have some scaling laws on that?
Very nice and timely work.
Where can one read in more details about your toy experiment setup and the large scale one? Like how did you define the causality score and why it has been concluded that causal confusion reduces by depth and more data? Do you have some scaling laws on that?
thanks! You can find the details in this paper https://arxiv.org/abs/2601.04575