Quantization Operator

Quantization is a model compression technique that reduces the precision of numerical representations in neural networks, aiming to decrease computational costs and memory footprint while preserving model accuracy. Current research focuses on applying quantization to various deep learning architectures, including Vision Transformers (ViTs), large language models (LLMs), and diffusion models, often employing post-training quantization (PTQ) methods to avoid retraining the entire model. This work is significant because it enables the deployment of large, computationally expensive models on resource-constrained devices, impacting fields like healthcare, edge computing, and natural language processing by making advanced AI more accessible and efficient.

Papers

August 12, 2024

Computability of Classification and Deep Learning: From Theoretical Limits to Practical Feasibility through Quantization
Holger Boche, Vit Fojtik, Adalbert Fono, Gitta Kutyniok
Deep Learning Classification Code Quantization Operator Deep Learning Application Fundamental Limit Banach Mazur Computable Computational Limit

August 7, 2024

Speaker Adaptation for Quantised End-to-End ASR Models
Qiuming Zhao, Guangzhi Sun, Chao Zhang, Mingxing Xu, Thomas Fang Zheng
Automatic Speech Recognition End to End Quantization Operator Automatic Speech Recognition Model Speaker Adaptation

August 6, 2024

July 29, 2024

MimiQ: Low-Bit Data-Free Quantization of Vision Transformers with Encouraging Inter-Head Attention Similarity
Kanghyun Choi, Hye Yoon Lee, Dain Kwon, SunJong Park, Kyuyeun Kim, Noseong Park, Jinho Lee
Vision Transformer Quantization Operator Low Bit Multi Head Attention Mechanism Data Free Quantization Attention Distillation Vision Transformer Quantization

July 20, 2024

MetaAug: Meta-Data Augmentation for Post-Training Quantization
Cuong Pham, Hoang Anh Dung, Cuong C. Nguyen, Trung Le, Dinh Phung, Gustavo Carneiro, Thanh-Toan Do
Supervised ImageNet Quantization Operator Post Training Quantization Calibration Data Meta Feature Augmentation

July 17, 2024

AdaLog: Post-Training Quantization for Vision Transformers with Adaptive Logarithm Quantizer
Zhuguanyu Wu, Jiaxin Chen, Hanwen Zhong, Di Huang, Yunhong Wang
Vision Transformer Quantization Operator Post Training Quantization Computer Vision Community Uniform Quantization

July 16, 2024

July 9, 2024

ERQ: Error Reduction for Post-Training Quantization of Vision Transformers
Yunshan Zhong, Jiawei Hu, You Huang, Yuxin Zhang, Rongrong Ji
Vision Transformer Quantization Operator Post Training Quantization Quantization Error Weight Quantization Activation Quantization Error Mitigation

July 5, 2024

The Impact of Quantization and Pruning on Deep Reinforcement Learning Models
Heng Lu, Mehdi Alemi, Reza Rawassizadeh
Deep Reinforcement Learning Global Impact Model Compression Quantization Operator Edge Pruning Neural Network Compression

July 3, 2024

June 25, 2024

June 20, 2024

Predicting Probabilities of Error to Combine Quantization and Early Exiting: QuEE
Florence Regol, Joud Chataoui, Bertrand Charpentier, Mark Coates, Pablo Piantanida, Stephan Gunnemann
Quantization Operator Error Feedback Dynamic Network Early eXit Post Training Predictive Probability Input Compression

June 19, 2024

SDQ: Sparse Decomposed Quantization for LLM Inference
Geonhwa Jeong, Po-An Tsai, Stephen W. Keckler, Tushar Krishna
Large Language Model Model Compression Quantization Operator LLM Inference Structured Sparsity

June 17, 2024

QTIP: Quantization with Trellises and Incoherence Processing
Albert Tseng, Qingyao Sun, David Hou, Christopher De Sa
Quantization Operator Post Training Quantization Vector Quantization Novel Encoding Extreme Quantization Tree Structure Quantization Performance Incoherence Processing

Quantization Operator

Papers

Computability of Classification and Deep Learning: From Theoretical Limits to Practical Feasibility through Quantization

Speaker Adaptation for Quantised End-to-End ASR Models

L3iTC at the FinLLM Challenge Task: Quantization for Financial Text Classification & Summarization

EC-Guide: A Comprehensive E-Commerce Guide for Instruction Tuning and Quantization

MimiQ: Low-Bit Data-Free Quantization of Vision Transformers with Encouraging Inter-Head Attention Similarity

MetaAug: Meta-Data Augmentation for Post-Training Quantization

AdaLog: Post-Training Quantization for Vision Transformers with Adaptive Logarithm Quantizer

Exploring Quantization for Efficient Pre-Training of Transformer Language Models

QVD: Post-training Quantization for Video Diffusion Models

LRQ: Optimizing Post-Training Quantization for Large Language Models by Learning Low-Rank Weight-Scaling Matrices

ERQ: Error Reduction for Post-Training Quantization of Vision Transformers

The Impact of Quantization and Pruning on Deep Reinforcement Learning Models

How Does Quantization Affect Multilingual LLMs?

Unified Anomaly Detection methods on Edge Device using Knowledge Distillation and Quantization

ADFQ-ViT: Activation-Distribution-Friendly Post-Training Quantization for Vision Transformers

CDQuant: Greedy Coordinate Descent for Accurate LLM Quantization

Layer-Wise Quantization: A Pragmatic and Effective Method for Quantizing LLMs Beyond Integer Bit-Levels

Predicting Probabilities of Error to Combine Quantization and Early Exiting: QuEE

SDQ: Sparse Decomposed Quantization for LLM Inference

QTIP: Quantization with Trellises and Incoherence Processing