Saved in:
Bibliographic Details
Main Authors: Tseng, Albert, Sun, Qingyao, Hou, David, De Sa, Christopher
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2406.11235
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916797180018688
author Tseng, Albert
Sun, Qingyao
Hou, David
De Sa, Christopher
author_facet Tseng, Albert
Sun, Qingyao
Hou, David
De Sa, Christopher
contents Post-training quantization (PTQ) reduces the memory footprint of LLMs by quantizing weights to low-precision datatypes. Since LLM inference is usually memory-bound, PTQ methods can improve inference throughput. Recent state-of-the-art PTQ approaches use vector quantization (VQ) to quantize multiple weights at once, which improves information utilization through better shaping. However, VQ requires a codebook with size exponential in the dimension. This limits current VQ-based PTQ works to low VQ dimensions ($\le 8$) that in turn limit quantization quality. Here, we introduce QTIP, which instead uses trellis coded quantization (TCQ) to achieve ultra-high-dimensional quantization. TCQ uses a stateful decoder that separates the codebook size from the bitrate and effective dimension. QTIP introduces a spectrum of lookup-only to computed lookup-free trellis codes designed for a hardware-efficient "bitshift" trellis structure; these codes achieve state-of-the-art results in both quantization quality and inference speed.
format Preprint
id arxiv_https___arxiv_org_abs_2406_11235
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle QTIP: Quantization with Trellises and Incoherence Processing
Tseng, Albert
Sun, Qingyao
Hou, David
De Sa, Christopher
Machine Learning
Post-training quantization (PTQ) reduces the memory footprint of LLMs by quantizing weights to low-precision datatypes. Since LLM inference is usually memory-bound, PTQ methods can improve inference throughput. Recent state-of-the-art PTQ approaches use vector quantization (VQ) to quantize multiple weights at once, which improves information utilization through better shaping. However, VQ requires a codebook with size exponential in the dimension. This limits current VQ-based PTQ works to low VQ dimensions ($\le 8$) that in turn limit quantization quality. Here, we introduce QTIP, which instead uses trellis coded quantization (TCQ) to achieve ultra-high-dimensional quantization. TCQ uses a stateful decoder that separates the codebook size from the bitrate and effective dimension. QTIP introduces a spectrum of lookup-only to computed lookup-free trellis codes designed for a hardware-efficient "bitshift" trellis structure; these codes achieve state-of-the-art results in both quantization quality and inference speed.
title QTIP: Quantization with Trellises and Incoherence Processing
topic Machine Learning
url https://arxiv.org/abs/2406.11235