Affine-Scaled Attention: Towards Flexible and Stable Transformer Attention

TL;DR

This paper introduces Affine-Scaled Attention, a modification to standard transformer attention that allows for input-dependent scaling and bias, improving training stability and performance in language models.

Contribution

The paper presents a novel attention mechanism that relaxes softmax normalization constraints, enabling better control over attention magnitudes and enhancing model training and performance.

Findings

Improved training stability in large-scale language models.

Enhanced downstream task performance.

Consistent benefits over standard softmax attention.

Abstract

Transformer attention is typically implemented using softmax normalization, which enforces attention weights with unit sum normalization. While effective in many settings, this constraint can limit flexibility in controlling attention magnitudes and may contribute to overly concentrated or unstable attention patterns during training. Prior work has explored modifications such as attention sinks or gating mechanisms, but these approaches provide only limited or indirect control over attention reweighting. We propose Affine-Scaled Attention, a simple extension to standard attention that introduces input-dependent scaling and a corresponding bias term applied to softmax-normalized attention weights. This design relaxes the strict normalization constraint while maintaining aggregation of value representations, allowing the model to adjust both the relative distribution and the scale of attention in a controlled manner. We empirically evaluate Affine-Scaled Attention in large-scale language model pretraining across multiple model sizes. Experimental results show consistent improvements in training stability, optimization behavior, and downstream task performance compared to standard softmax attention and attention sink baselines. These findings suggest that modest reweighting of attention outputs provides a practical and effective way to improve attention behavior in Transformer models.