Cuong Pham
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Cuong Pham has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
- Anh Dung Hoang
Department of Data Science and AI Department of Data Science and AI Speech and Signal Processing Department of Data Science and AI Department of Data Science and AI Department of Data Science and AI Centre for Vision Monash University Monash University University of Surrey Monash University Monash University Monash University
1 shared publication - Cuong Nguyen
University of Surrey
1 shared publication - Jianfei Cai
Monash University, Melbourne, VIC, Australia
1 shared publication Thanh-Toan Do
Department of Data Science and AI Department of Data Science and AI Speech and Signal Processing Department of Data Science and AI Department of Data Science and AI Department of Data Science and AI Centre for Vision Monash University Monash University University of Surrey Monash University Monash University Monash University
1 shared publication- Trung Le
Monash University, Melbourne, VIC, Australia
1 shared publication
Latest Publications
- Open Access
Rethinking Output Alignment For 1-bit Post-Training Quantization of Large Language Models
Large Language Models (LLMs) deliver strong performance across a wide range of NLP tasks, but their massive sizes hinder deployment on resource-constrained devices. To reduce their computational and memory burden, various compression techniques have been proposed, including quantization, pruning, and knowledge distillation. Among these, post-training quantization (PTQ) is widely adopted for its efficiency, as it requires no retraining and only a small dataset for calibration, enabling low-cost deployment. Recent advances for post-training quantization have demonstrated that even sub-4-bit methods can maintain most of the original model performance. However, 1-bit quantization that converts floating-point weights to \(\pm\)1, remains particularly challenging, as existing 1-bit PTQ methods often suffer from significant performance degradation compared to the full-precision models. Specifically, most of existing 1-bit PTQ approaches focus on weight alignment, aligning the full-precision model weights with those of the quantized models, rather than directly aligning their outputs. Although the output-matching approach objective is more intuitive and aligns with the quantization goal, naively applying it in 1-bit LLMs often leads to notable performance degradation. In this paper, we investigate why and under what conditions output-matching fails, in the context of 1-bit LLM quantization. Based on our findings, we propose a novel data-aware PTQ approach for 1-bit LLMs that explicitly accounts for activation error accumulation while keeping optimization efficient. Empirical experiments demonstrate that our solution consistently outperforms existing 1-bit PTQ methods with minimal overhead.Open AccessRethinking Output Alignment For 1-bit Post-Training Quantization of Large Language Models
Large Language Models (LLMs) deliver strong performance across a wide range of NLP tasks, but their massive sizes hinder deployment on resource-constrained devices. To reduce their computational and memory burden, various compression techniques have been proposed, including quantization, pruning, and knowledge distillation. Among these, post-training quantization (PTQ) is widely adopted for its efficiency, as it requires no retraining and only a small dataset for calibration, enabling low-cost deployment. Recent advances for post-training quantization have demonstrated that even sub-4-bit methods can maintain most of the original model performance. However, 1-bit quantization that converts floating-point weights to \(\pm\)1, remains particularly challenging, as existing 1-bit PTQ methods often suffer from significant performance degradation compared to the full-precision models. Specifically, most of existing 1-bit PTQ approaches focus on weight alignment, aligning the full-precision model weights with those of the quantized models, rather than directly aligning their outputs. Although the output-matching approach objective is more intuitive and aligns with the quantization goal, naively applying it in 1-bit LLMs often leads to notable performance degradation. In this paper, we investigate why and under what conditions output-matching fails, in the context of 1-bit LLM quantization. Based on our findings, we propose a novel data-aware PTQ approach for 1-bit LLMs that explicitly accounts for activation error accumulation while keeping optimization efficient. Empirical experiments demonstrate that our solution consistently outperforms existing 1-bit PTQ methods with minimal overhead.