E-PMQ: Expert-Guided Post-Merge Quantization with Merged-Weight Anchoring

TL;DR

E-PMQ introduces an expert-guided post-merge quantization framework that stabilizes and improves low-bit model deployment by leveraging source expert weights and merged-weight anchoring.

Contribution

The paper proposes E-PMQ, a novel method for post-merge quantization that enhances low-bit model accuracy by mitigating merging and quantization deviations using expert guidance and weight anchoring.

Findings

E-PMQ improves 4-bit GPTQ accuracy from 65.0% to 73.6% on CLIP-ViT-B/32 eight-task merging.

E-PMQ increases GPTQ accuracy from 34.8% to 76.7% on 20-task CLIP-ViT-L/14.

E-PMQ achieves 83.34% accuracy on FLAN-T5-base GLUE, outperforming prior methods.

Abstract

Low-resource deployment constraints have made model quantization essential for deploying neural networks while preserving performance. Meanwhile, model merging has become an increasingly practical low-resource strategy for integrating multiple task- or domain-specialized experts into a single model without joint training or multi-model serving. Together, quantization and model merging enable an efficient low-resource deployment pipeline by integrating multiple experts into one low-bit model. We formulate this setting as Post-Merge Quantization (PMQ). We show that directly applying post-training quantization (PTQ) to a merged model is unreliable because two distinct deviations are coupled: the quantization deviation introduced by low-bit reconstruction and the expert-relative merging deviation inherited from model merging. To mitigate these deviations, we propose E-PMQ, an expert-guided PMQ framework that uses source expert weights to provide expert- guided output targets during layer-wise calibration, together with merged-weight anchoring to stabilize the calibration and preserve the integrated behavior of the merged model. On CLIP-ViT-B/32 eight-task merging, E-PMQ improves 4-bit GPTQ from 65.0% to 73.6% under Task Arithmetic and from 69.1% to 74.8% under TIES-Merging. On harder settings, E-PMQ improves GPTQ from 34.8% to 76.7% on 20-task CLIP-ViT-L/14 and from 78.26% to 83.34% on FLAN-T5- base GLUE. These results demonstrate that E-PMQ enables effective post-merge quantization and low-bit deployment.