Agile-Quant: Activation-Guided Quantization for Faster Inference of LLMs on the Edge

TL;DR

Agile-Quant introduces an activation-guided quantization framework and accelerator for large language models, enabling faster on-device inference on edge devices with minimal performance loss through 4-bit and 8-bit quantization.

Contribution

The paper presents a novel activation-aware quantization method and an efficient accelerator for LLMs, achieving significant speedups on edge devices while maintaining accuracy.

Findings

Achieves up to 2.55x speedup on edge devices with 4-bit and 8-bit quantization.

Maintains task performance comparable to weight-only quantization methods.

Successfully applies quantization to LLaMA, OPT, and BLOOM models.

Abstract

Large Language Models (LLMs) stand out for their impressive performance in intricate language modeling tasks. However, their demanding computational and memory needs pose obstacles for broad use on edge devices. Quantization is then introduced to boost LLMs' on-device efficiency. Recent works show that 8-bit or lower weight quantization is feasible with minimal impact on end-to-end task performance, while the activation is still not quantized. On the other hand, mainstream commodity edge devices still struggle to execute these sub-8-bit quantized networks effectively. In this paper, we propose Agile-Quant, an activation-guided quantization framework for popular Large Language Models (LLMs), and implement an end-to-end accelerator on multiple edge devices for faster inference. Considering the hardware profiling and activation analysis, we first introduce a basic activation quantization strategy to balance the trade-off of task performance and real inference speed. Then we leverage the activation-aware token pruning technique to reduce the outliers and the adverse impact on attentivity. Ultimately, we utilize the SIMD-based 4-bit multiplier and our efficient TRIP matrix multiplication to implement the accelerator for LLMs on the edge. We apply our framework on different scales of LLMs including LLaMA, OPT, and BLOOM with 4-bit or 8-bit for the activation and 4-bit for the weight quantization. Experiments show that Agile-Quant achieves simultaneous quantization of model weights and activations while maintaining task performance comparable to existing weight-only quantization methods. Moreover, in the 8- and 4-bit scenario, Agile-Quant achieves an on-device speedup of up to 2.55x compared to its FP16 counterparts across multiple edge devices, marking a pioneering advancement in this domain. Code: https://github.com/shawnricecake/agile-quant