Saved in:
Bibliographic Details
Main Authors: Ashkboos, Saleh, Verhoef, Bram, Hoefler, Torsten, Eleftheriou, Evangelos, Dazzi, Martino
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.11038
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910702416953344
author Ashkboos, Saleh
Verhoef, Bram
Hoefler, Torsten
Eleftheriou, Evangelos
Dazzi, Martino
author_facet Ashkboos, Saleh
Verhoef, Bram
Hoefler, Torsten
Eleftheriou, Evangelos
Dazzi, Martino
contents Quantization-aware training (QAT) schemes have been shown to achieve near-full precision accuracy. They accomplish this by training a quantized model for multiple epochs. This is computationally expensive, mainly because of the full precision backward pass. On the other hand, post-training quantization (PTQ) schemes do not involve training and are therefore computationally cheap, but they usually result in a significant accuracy drop. We address these challenges by proposing EfQAT, which generalizes both schemes by optimizing only a subset of the parameters of a quantized model. EfQAT starts by applying a PTQ scheme to a pre-trained model and only updates the most critical network parameters while freezing the rest, accelerating the backward pass. We demonstrate the effectiveness of EfQAT on various CNNs and Transformer-based models using different GPUs. Specifically, we show that EfQAT is significantly more accurate than PTQ with little extra compute. Furthermore, EfQAT can accelerate the QAT backward pass between 1.44-1.64x while retaining most accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2411_11038
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle EfQAT: An Efficient Framework for Quantization-Aware Training
Ashkboos, Saleh
Verhoef, Bram
Hoefler, Torsten
Eleftheriou, Evangelos
Dazzi, Martino
Machine Learning
Quantization-aware training (QAT) schemes have been shown to achieve near-full precision accuracy. They accomplish this by training a quantized model for multiple epochs. This is computationally expensive, mainly because of the full precision backward pass. On the other hand, post-training quantization (PTQ) schemes do not involve training and are therefore computationally cheap, but they usually result in a significant accuracy drop. We address these challenges by proposing EfQAT, which generalizes both schemes by optimizing only a subset of the parameters of a quantized model. EfQAT starts by applying a PTQ scheme to a pre-trained model and only updates the most critical network parameters while freezing the rest, accelerating the backward pass. We demonstrate the effectiveness of EfQAT on various CNNs and Transformer-based models using different GPUs. Specifically, we show that EfQAT is significantly more accurate than PTQ with little extra compute. Furthermore, EfQAT can accelerate the QAT backward pass between 1.44-1.64x while retaining most accuracy.
title EfQAT: An Efficient Framework for Quantization-Aware Training
topic Machine Learning
url https://arxiv.org/abs/2411.11038