M-ANT: Efficient Low-bit Group Quantization for LLMs via Mathematically Adaptive Numerical Type

TL;DR

MANT introduces a mathematically adaptive low-bit quantization method for LLMs, enabling more flexible and efficient compression of model weights and caches, leading to significant speed and energy improvements.

Contribution

The paper proposes MANT, a novel adaptive numeric type and supporting framework for group-wise quantization, addressing distribution diversity and real-time processing challenges in LLM deployment.

Findings

Achieves nearly 3x speedup over state-of-the-art accelerators.

Reduces energy consumption by over 2.8x.

Effectively unifies weight and cache quantization processes.

Abstract

Large language models (LLMs) are one of the most important killer computer applications. The recent algorithmic advancement proposes a fine-grained group-wise quantization for LLMs, which treats a small set (e.g., 64) of values in a tensor as a compression unit. It effectively preserves the model accuracy without retraining, and has become the standard approach to efficiently deploy LLMs. On the other hand, there are works that propose various adaptive data types to better adapt to different distributions and further reduce the required bit length for LLMs. In this work, our detailed analysis unveils a key finding that while different tensors exhibit similar distributions, small groups can have markedly different distributions. As such, the group-level diversity requires a new level of adaptivity for which existing adaptive data types fail to provide. In this paper, we propose MANT, a mathematically adaptive numeric type, featuring a more flexible encoding paradigm with a wider range of data distribution and more efficient decodingcomputation fusion mechanism to address these challenges. Based on MANT, we develop a supporting framework to assign the appropriate data type for each group adaptively. Meanwhile, the dynamically generated Key-Value (KV) caches in LLMs introduce further complexity for real-time quantization. To tackle this, we propose an efficient real-time quantization mechanism. Besides, we implement a specific processing element (PE) to efficiently support MANT and incorporate a real-time quantization unit. By integrating these components into a systolic array, MANT unifies the group-wise weight and KV cache quantization and addresses the associated challenges. Our evaluation shows achieving, on average, 2.99x (up to 4.46x) speedup and 2.81x (up to 4.10x) energy reduction to the state-of-the-art LLM accelerator.