SKIM: Any-bit Quantization Pushing The Limits of Post-Training Quantization

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Bai, Runsheng, Liu, Bo, Liu, Qiang
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866929619499745280
author Bai, Runsheng
Liu, Bo
Liu, Qiang
author_facet Bai, Runsheng
Liu, Bo
Liu, Qiang
contents Large Language Models (LLMs) exhibit impressive performance across various tasks, but deploying them for inference poses challenges. Their high resource demands often necessitate complex, costly multi-GPU pipelines, or the use of smaller, less capable models. While quantization offers a promising solution utilizing lower precision for model storage, existing methods frequently experience significant performance drops at lower precision levels. Additionally, they typically provide only a limited set of solutions at specific bit levels, many of which are extensively manually tuned. To address these challenges, we propose a new method called SKIM: Scaled K-means clustering wIth Mixed precision. Our approach introduces two novel techniques: 1. A greedy algorithm to solve approximately optimal bit allocation across weight channels, and 2. A trainable scaling vector for non-differentiable K-means clustering. These techniques substantially improve performance and can be adapted to any given bit. Notably, in terms of model perplexity, our method narrows the gap between 3-bit quantized LLaMA models and their full precision counterparts by 16.3% on average.
format Preprint
id arxiv_https___arxiv_org_abs_2412_04180
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SKIM: Any-bit Quantization Pushing The Limits of Post-Training Quantization
Bai, Runsheng
Liu, Bo
Liu, Qiang
Machine Learning
Large Language Models (LLMs) exhibit impressive performance across various tasks, but deploying them for inference poses challenges. Their high resource demands often necessitate complex, costly multi-GPU pipelines, or the use of smaller, less capable models. While quantization offers a promising solution utilizing lower precision for model storage, existing methods frequently experience significant performance drops at lower precision levels. Additionally, they typically provide only a limited set of solutions at specific bit levels, many of which are extensively manually tuned. To address these challenges, we propose a new method called SKIM: Scaled K-means clustering wIth Mixed precision. Our approach introduces two novel techniques: 1. A greedy algorithm to solve approximately optimal bit allocation across weight channels, and 2. A trainable scaling vector for non-differentiable K-means clustering. These techniques substantially improve performance and can be adapted to any given bit. Notably, in terms of model perplexity, our method narrows the gap between 3-bit quantized LLaMA models and their full precision counterparts by 16.3% on average.
title SKIM: Any-bit Quantization Pushing The Limits of Post-Training Quantization
topic Machine Learning
url https://arxiv.org/abs/2412.04180