AI Paper Atlas

Research map / Architectures and efficiency / Quantization

LiftQuant: Continuous Bit-Width LLM via Dimensional Lifting and Projection

Liulu He, Xuan Ang Liu, Juntao Liu, Taolue Feng, Ting Lu, Chunsheng Gan, ZHIYV PENG, Yuan Du, Huanrui Yang, Yijiang Liu, Li Du

ICML 2026 · Architectures and efficiency › Quantization · approach: KV cache quantization · 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
add scaling s1 (ablation)LiftQuant on Llama-2-7B · WikiText-2perplexity: 8.76 -> 8.28Improved
add scaling s2 (ablation)LiftQuant on Llama-2-7B · WikiText-2perplexity: 8.28 -> 7.77Improved
add lifted projection matrix M instead of 2-bit uniform quantizer (ablation)LiftQuant on Llama-2-7B · WikiText-2perplexity: 7.77 -> 6.79Improved
add end-to-end fine-tuning (ablation)LiftQuant on Llama-2-7B · WikiText-2perplexity: 6.79 -> 6.53Improved
remove orthogonal mixing matrices P1 and P2 (ablation)LiftQuant whitening · Llama-2-7Btraining stabilityHurt
increase bit-width from 2.0 to 2.14 (LQ-32/16 to LQ-30/14)LiftQuant · Llama-2-7B WikiText-2perplexity: 6.53 -> 6.30Improved
increase bit-width to 2.4 (LQ-24/10) from 2-bit LQ-32/16LiftQuant · Llama-3-70B WikiText-2perplexity: 4.69 -> 4.10Improved
increase bit-width to 2.4 (LQ-24/10) from 2-bit LQ-32/16LiftQuant · Llama-2-70B WikiText-2perplexity: 4.08 -> 3.78Improved
increase lifted dimension from LQ-16/8 to LQ-32/16 at 2 bitsLiftQuant · Standard Gaussian / Llama-2-7Bperplexity: 6.60 -> 6.53Improved
increase search dimension gap (LQ-32/16 vs LQ-16/8)LiftQuant · nearest-neighbor search timesearch time: <<0.1S -> 4SHurt
use LiftQuant 2-bit instead of QTIPLiftQuant LQ-32/16 · Llama-2-7B WikiText-2perplexity: 6.28 -> 6.52Hurt
use LiftQuant 2-bit instead of QTIPLiftQuant LQ-32/16 · Llama-3-70B WikiText-2perplexity: 4.97 -> 4.69Improved

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