HARP: Hesitation-Aware Reframing in Transformer Inference Pass

TL;DR

HARP is a model-agnostic, training-free method that adaptively adds computation during Transformer inference by mimicking human hesitation, leading to improved performance and faster inference times.

Contribution

HARP introduces a novel hesitation-aware reframing technique that enhances Transformer inference without retraining or complex modifications.

Findings

Up to +5.16% performance improvement on downstream tasks

Inference times are twice as fast as beam search

Applicable across various models and tasks

Abstract

This paper aims to improve the performance of large language models by addressing the variable computational demands in inference steps, where some tokens require more computational resources than others. We present HARP, a simple modification to "off-the-shelf" Transformer forward pass. Drawing from hesitation and the framing effect in decision-making, HARP selectively applies additional computation when the model encounters uncertainty during token generation. Our method mimics human cognitive processes by pausing at difficult decision points and reframing inputs for a different perspective. Unlike other approaches, HARP is model-agnostic, training-free, and easy to implement. We evaluate our method across various downstream tasks and model sizes, demonstrating performance improvements up to +5.16%. Notably, HARP achieves these gains while maintaining inference times twice faster than beam search. Simple and yet with significant gains, HARP provides insights into the potential of adaptive computation for enhancing the performance of Transformer-based language models.