PolarQuant: Leveraging Polar Transformation for Efficient Key Cache Quantization and Decoding Acceleration

TL;DR

PolarQuant introduces a polar coordinate-based quantization method for key caches in large language models, effectively handling outliers and significantly accelerating decoding without sacrificing model performance.

Contribution

The paper presents PolarQuant, a novel 2D vector grouping and polar coordinate encoding approach for efficient key cache quantization in large language models.

Findings

Reduces memory usage of KV caches in large models.

Speeds up decoding by transforming inner products into table lookups.

Maintains model performance comparable to full-precision models.

Abstract

The KV cache in large language models is a dominant factor in memory usage, limiting their broader applicability. Quantizing the cache to lower bit widths is an effective way to reduce computational costs; however, previous methods struggle with quantizing key vectors due to outliers, resulting in excessive overhead. We propose a novel quantization approach called PolarQuant, which efficiently addresses the outlier challenge. We observe that outliers typically appear in only one of two dimensions, which are rotated together by a specific angle when rotary position embeddings are applied. When represented as two-dimensional vectors, these dimensions exhibit well-structured patterns, with radii and angles smoothly distributed in polar coordinates. This alleviates the challenge of outliers on per-channel quantization, making them well-suited for quantization. Thus, PolarQuant divides key vectors into groups of two-dimensional sub-vectors, encoding them as the corresponding quantized radius and the polar angle, rather than quantizing original key vectors directly. PolarQuant achieves the superior efficiency in KV cache quantization and accelerates the decoding process by turning the query-key inner product into a table lookup, all while maintaining the downstream performance of full-precision models.