Does Transformer Interpretability Transfer to RNNs?

arXiv:2404.05971v1 Announce Type: new
Abstract: Recent advances in recurrent neural network architectures, such as Mamba and RWKV, have enabled RNNs to match or exceed the performance of equal-size transformers in terms of language modeling perplexity and downstream evaluations, suggesting that future systems may be built on completely new architectures. In this paper, we examine if selected interpretability methods originally designed for transformer language models will transfer to these up-and-coming recurrent architectures. Specifically, we focus on steering model outputs via contrastive …

abstract advances architectures arxiv cs.ai cs.cl cs.lg future interpretability language mamba match modeling network neural network paper performance perplexity recurrent neural network rwkv systems terms transfer transformer transformers type