AI Paper Atlas

Research map / Architectures and efficiency / Quantization

BPDQ: Bit-Plane Decomposition Quantization on a Variable Grid for Large Language Models

Junyu Chen, Jungang Li, Jing Xiong, Wenjie Wang, Qingyao Yang, He Xiao, Zhen Li, Taiqiang Wu, Mengzhao Chen, Zhen Peng, Chaofan Tao, Long Shi and 2 more

ICML 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 a variable bit-plane grid instead of a fixed uniform grid (BPDQ vs GPTQ)2-bit quantization · Qwen2.5-72B GSM8Kaccuracy: 63.46% -> 87.72%Improved
use BPDQ instead of GPTQ at 2-bitW2 quantization · Qwen3-32B GSM8Kaccuracy: 44.20% -> 80.89%Improved
use BPDQ instead of GPTQ at 2-bitW2 quantization · Ministral3-8B GSM8Kaccuracy: 12.36% -> 42.46%Improved
use BPDQ instead of AWQ at 2-bitW2 quantization · Qwen2.5-72B GSM8Kaccuracy: 0.00% -> 87.72%Improved
use outlier protection (AWQ) alone at 2-bitAWQ · 2-bit reasoningaccuracy: 0.00%Hurt
increase group size from 64 to 256 in BPDQ 2-bit (ablation)BPDQ · Qwen2.5-72B GSM8Kaccuracy: 87.72% -> 83.85%Hurt
BPDQ vs VPTQ at 2-bitBPDQ · Qwen2.5-7B GSM8Kaccuracy: 67.63% -> 44.50%Hurt
BPDQ vs GPTQ at 2-bit on small modelBPDQ · Qwen3-0.6B GSM8Kaccuracy: 0.00% -> 0.15%No clear effect
BPDQ vs GPTQ at 4-bitBPDQ · Qwen3-4B GSM8Kaccuracy: 85.44% -> 81.50%Hurt
BPDQ vs GPTQ quantization timeBPDQ · Qwen2.5-7B W4minutes: 16 -> 47Hurt
BPDQ vs GPTQ decoding latency 2-bitBPDQ · Qwen2.5-7B W2latency (ms): 33.91 -> 18.09Improved
fixed-grid GPTQ suppresses outliers at 2-bitGPTQ-W2-G32 · Qwen2.5-7BDiagR change: -32.89%Hurt

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