Efficient Sample-Specific Encoder Perturbations

arXiv:2405.01601v1 Announce Type: cross
Abstract: Encoder-decoder foundation models have displayed state-of-the-art performance on a range of autoregressive sequence tasks. This paper proposes a simple and lightweight modification to such systems to control the behaviour according to a specific attribute of interest. This paper proposes a novel inference-efficient approach to modifying the behaviour of an encoder-decoder system according to a specific attribute of interest. Specifically, we show that a small proxy network can be used to find a sample-by-sample perturbation of …

abstract art arxiv autoregressive control cs.cl cs.lg decoder encoder encoder-decoder foundation inference novel paper performance sample simple state systems tasks type