Summary of Fp=xint:a Low-bit Series Expansion Algorithm For Post-training Quantization, by Boyang Zhang et al.
FP=xINT:A Low-Bit Series Expansion Algorithm for Post-Training Quantization
by Boyang Zhang, Daning Cheng, Yunquan Zhang, Fangmin Liu
First submitted to arxiv on: 9 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces a novel method called Post-Training Quantization (PTQ) that converts pre-trained Full-Precision (FP) models into quantized versions without training. Existing methods reduce size and computational costs but degrade performance at extremely low settings due to quantization noise. The authors propose a deep model series expansion framework that enables rapid and accurate approximation of unquantized models without calibration sets or fine-tuning. This is the first use of series expansion for neural network quantization. The method expands the FP model into multiple low-bit basis models, ensuring accurate quantization through low-bit basis model expansions at different granularities (tensor, layer, model). Experiments show that this algorithm achieves state-of-the-art performance in low-bit settings, surpassing original accuracy with 4-bit quantization of ResNet-50 reaching 77.03%. The code will be made public. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary PTQ converts pre-trained FP models into quantized versions without training. Existing methods reduce size and computational costs but degrade performance at extremely low settings due to quantization noise. A new method uses series expansion to approximate unquantized models, achieving rapid and accurate results. This is the first use of series expansion for neural network quantization. |
Keywords
» Artificial intelligence » Fine tuning » Neural network » Precision » Quantization » Resnet
