GAMMA: Global Bit Allocation for Mixed-Precision Models under Arbitrary Budgets

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yao, Zhangyang, Zhao, Haiyan, Wang, Haoyu, Huang, Tianbo, Zhang, Lihua, Han, Xu
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918509340000256
author Yao, Zhangyang
Zhao, Haiyan
Wang, Haoyu
Huang, Tianbo
Zhang, Lihua
Han, Xu
author_facet Yao, Zhangyang
Zhao, Haiyan
Wang, Haoyu
Huang, Tianbo
Zhang, Lihua
Han, Xu
contents Mixed-precision quantization improves the budget--accuracy trade-off for large language models (LLMs) by allocating more bits to sensitive modules. However, automating this allocation at LLM scale faces a unique combination of constraints: learnable approaches require quantization-aware training, which is infeasible for billion-parameter models; training-free alternatives rely on static proxy metrics that miss cross-module interactions and must be recomputed per target budget; and search-based methods are expensive without guaranteeing exact budget compliance. We propose GAMMA, a quantizer-agnostic framework that learns module-wise precision preferences entirely within a post-training pipeline. GAMMA optimizes a teacher-forced hidden-state reconstruction objective under an augmented Lagrangian constraint, and projects the learned preferences into exact budget-feasible discrete assignments via integer programming. A key property is score reuse: because the learned preferences encode a stable sensitivity ranking rather than budget-specific weights, a single training run serves arbitrary deployment targets by re-solving only the integer program, reducing per-budget adaptation from hours to a few minutes. Across Llama and Qwen models (8B--32B), GAMMA outperforms both fixed-precision baselines (up to +12.99 Avg.) and search-based mixed-precision methods (up to +7.00 Avg.), and can match fixed 3-bit quality at 2.5-bit average precision, enabling deployment at substantially smaller memory footprints.
format Preprint
id arxiv_https___arxiv_org_abs_2605_18475
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle GAMMA: Global Bit Allocation for Mixed-Precision Models under Arbitrary Budgets
Yao, Zhangyang
Zhao, Haiyan
Wang, Haoyu
Huang, Tianbo
Zhang, Lihua
Han, Xu
Machine Learning
Artificial Intelligence
Mixed-precision quantization improves the budget--accuracy trade-off for large language models (LLMs) by allocating more bits to sensitive modules. However, automating this allocation at LLM scale faces a unique combination of constraints: learnable approaches require quantization-aware training, which is infeasible for billion-parameter models; training-free alternatives rely on static proxy metrics that miss cross-module interactions and must be recomputed per target budget; and search-based methods are expensive without guaranteeing exact budget compliance. We propose GAMMA, a quantizer-agnostic framework that learns module-wise precision preferences entirely within a post-training pipeline. GAMMA optimizes a teacher-forced hidden-state reconstruction objective under an augmented Lagrangian constraint, and projects the learned preferences into exact budget-feasible discrete assignments via integer programming. A key property is score reuse: because the learned preferences encode a stable sensitivity ranking rather than budget-specific weights, a single training run serves arbitrary deployment targets by re-solving only the integer program, reducing per-budget adaptation from hours to a few minutes. Across Llama and Qwen models (8B--32B), GAMMA outperforms both fixed-precision baselines (up to +12.99 Avg.) and search-based mixed-precision methods (up to +7.00 Avg.), and can match fixed 3-bit quality at 2.5-bit average precision, enabling deployment at substantially smaller memory footprints.
title GAMMA: Global Bit Allocation for Mixed-Precision Models under Arbitrary Budgets
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2605.18475