AI Paper Atlas

Research map / Architectures and efficiency / Quantization

Stabilizing Ultra-Low-Bit Quantization of Multimodal LLMs via Global Bit Allocation

Guilin Li, Yuexiao Ma, Yue Zhang, Xinxiong Wu, Jiaqi Zhou, Qingheng Zhang, Yan Zhang, Fei Chao, Xiawu Zheng, Rongrong Ji

ECCV 2026 · Architectures and efficiency › Quantization · approach: Post-training quantization of LLMs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use GloBitQ mixed-precision (3.01-bit) instead of uniform per-channel GPTQ/VLMQ at 3-bitQwen2-VL-7B-Instruct · 6-benchmark average, 3-bitaverage score: 64.74 -> 81.70 (VLMQ), full precision 83.76Improved
use GloBitQ instead of VLMQ at 3-bitQwen2.5-VL-7B-Instruct · 6-benchmark average, 3-bitaverage score: 77.49 -> 78.3Mixed
use GloBitQ instead of VLMQ at 3-bitLLaVA-OneVision · 6-benchmark average, 3-bitaverage score: 72.17 -> 76.66Improved
use GloBitQ per-channel 2.01-bit instead of VLMQ group-128 at 2-bitQwen2-VL-7B-Instruct · 6-benchmark average, 2-bitaverage score: 64.0 -> 66.6Improved
use GloBitQ per-channel 2.01-bit instead of VLMQ group-128Qwen2.5-VL-7B-Instruct · 6-benchmark average, 2-bitaverage score: 64.6 -> 66.4Improved
use GloBitQ per-channel 2.01-bit instead of VLMQ group-128LLaVA-OneVision · 6-benchmark average, 2-bitaverage score: 61.5 -> 60.8No clear effect
raise average bit-width from 2.01 to 2.05 (ablation)GloBitQ · LLaVA-OneVision 2-bit, 6-benchmark averageaverage score: 60.8 -> 63.1Improved
add global importance-balanced mixed precision (2.01 bit) over uniform 2-bit FlatQuant/GPTQ (ablation)Qwen2-VL-7B-Instruct · 6-benchmark average, 2-bitaverage score: 53.5 -> 66.6 (GPTQ), 51.1 -> 66.6 (FlatQuant)Improved
use per-channel GPTQ at 2.01-bit instead of GloBitQQwen2-VL, Qwen2.5-VL, LLaVA-OneVision · 2-bitaverage score: NAN (collapse)Hurt
set balance factor alpha to 0.5 rather than 0.2-0.3 or 0.7-0.8 (ablation)GloBitQ · LLaVA-OneVision-Int2.01average scoreImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy