Attention Entropy is a Key Factor: An Analysis of Parallel Context Encoding with Full-attention-based Pre-trained Language Models

TL;DR

This paper analyzes how high attention entropy affects parallel context encoding in large language models and proposes methods to reduce entropy, improving performance and efficiency in context modeling.

Contribution

It identifies attention entropy as a key factor in parallel encoding performance and introduces simple techniques to mitigate its effects.

Findings

Reducing attention entropy improves model performance.

Parallel context encoding can be optimized by attention sinks and selective mechanisms.

High attention entropy correlates with performance degradation.

Abstract

Large language models have shown remarkable performance across a wide range of language tasks, owing to their exceptional capabilities in context modeling. The most commonly used method of context modeling is full self-attention, as seen in standard decoder-only Transformers. Although powerful, this method can be inefficient for long sequences and may overlook inherent input structures. To address these problems, an alternative approach is parallel context encoding, which splits the context into sub-pieces and encodes them parallelly. Because parallel patterns are not encountered during training, naively applying parallel encoding leads to performance degradation. However, the underlying reasons and potential mitigations are unclear. In this work, we provide a detailed analysis of this issue and identify that unusually high attention entropy can be a key factor. Furthermore, we adopt two straightforward methods to reduce attention entropy by incorporating attention sinks and selective mechanisms. Experiments on various tasks reveal that these methods effectively lower irregular attention entropy and narrow performance gaps. We hope this study can illuminate ways to enhance context modeling mechanisms.