analiza aiqbench
SAGE-PTQ addresses hidden scaling overheads in ultra-low-bit quantization by leveraging graph-guided saliency analysis to dynamically allocate precision between salient and unsalient weights. This approach significantly reduces memory footprint and inference latency compared to existing methods like BiLLM, achieving near-optimal perplexity with minimal scaling bits. The technique offers a practical pathway for efficient LLM deployment by eliminating rigid heuristics that previously inflated computational costs.
Minimizing the Hidden Cost of Scales: Graph-Guided Ultra-Low-Bit Quantization for Large Language Models
arXiv:2606.05429v1 Announce Type: new Abstract: Post-training quantization (PTQ) is critical for the efficient deployment of large language models (LLMs). Recent ultra-low-bit PTQ methods rely on rigid weight-saliency assumptions or position heuristics, introducing substantial hidden scaling overhead. We propose SAGE-PTQ (Saliency-Aware Graph-guided Efficient PTQ), a novel ultra-low-bit quantization framework for LLMs that minimizes hidden scaling cost. SAGE-PTQ separates salient and unsalient weights using distributional statistics, then models subsampled unsalient weights as a sparse graph to estimate the optimal number of groups per layer. SAGE-PTQ applies dual-mode quantization, assigning multi-bit precision to salient weights and binarizing unsalient weights. To reduce scaling overhead, SAGE-PTQ uses one per-channel scale for salient weights and one scalar per unsalient group. Finally, SAGE-PTQ implements adaptive saliency thresholding to select the optimal saliency ratio per matrix. SAGE-PTQ achieves 1.03 weight bits and only 0.004 scaling bits per matrix on average, outperforming state-of-the-art methods such as BiLLM and PB-LLM. On LLaMA-3-8B, SAGE-PTQ achieves 6.74 WikiText2 perplexity, compared to 55.8 for BiLLM, while using less than 50% of BiLLM's GPU memory. On LLaMA-2-70B, SAGE-PTQ provides 1.5x faster decoding on one NVIDIA L40 GPU, demonstrating practical inference efficiency.
Abstract:Post-training quantization (PTQ) is critical for the efficient deployment of large language models (LLMs). Recent ultra-low-bit PTQ methods rely on rigid weight-saliency assumptions or position heuristics, introducing substantial hidden scaling overhead. We propose SAGE-PTQ (Saliency-Aware Graph-guided Efficient PTQ), a novel ultra-low-bit quantization framework for LLMs that minimizes hidden scaling cost. SAGE-PTQ separates salient and unsalient weights using distributional statistics, then models subsampled unsalient weights as a sparse graph to estimate the optimal number of groups per layer. SAGE-PTQ applies dual-mode quantization, assigning multi-bit precision to salient weights and binarizing unsalient weights. To reduce scaling overhead, SAGE-PTQ uses one per-channel scale for salient weights and one scalar per unsalient group. Finally, SAGE-PTQ implements adaptive saliency thresholding to select the optimal saliency ratio per matrix. SAGE-PTQ achieves 1.03 weight bits and only 0.004 scaling bits per matrix on average, outperforming state-of-the-art methods such as BiLLM and PB-LLM. On LLaMA-3-8B, SAGE-PTQ achieves 6.74 WikiText2 perplexity, compared to 55.8 for BiLLM, while using less than 50% of BiLLM's GPU memory. On LLaMA-2-70B, SAGE-PTQ provides 1.5x faster decoding on one NVIDIA L40 GPU, demonstrating practical inference efficiency.
Preprint. 18 pages, 10 figures, 7 tables, including appendix
Artificial Intelligence (cs.AI)
arXiv:2606.05429 [cs.AI]
(or arXiv:2606.05429v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2606.05429 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 285 cuvinte
Subiecte conexe
GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model
Meta Engineering · 3 aug. 2026
Deploying Qwen3.8-2.4T-A95B on Amazon SageMaker HyperPod with vLLM
AWS · 9 sept. 2026
NVIDIA Blackwell Tops MLPerf Training 6.0 with Industry-Leading Scale and Performance
NVIDIA · 16 iun. 2026
News
Quantum Computing Report · 25 iul. 2026