AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Efficient-LVSM: Faster, Cheaper, and Better Large View Synthesis Model via Decoupled Co-Refinement Attention

Xiaosong Jia, Yihang Sun, Junqi You, Songbur Wong, Zichen Zou, Junchi Yan, Zuxuan Wu, Yu-Gang Jiang

ICLR 2026 · Architectures and efficiency › Attention and transformers · approach: Long-context KV and memory attention · 5 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 self-then-cross attention in decoder (ablation)Efficient-LVSM 6+6 · RealEstate10KPSNR (dB): 24.18 -> 24.97Improved
add encoder-decoder co-refinement (ablation)Efficient-LVSM 6+6 · RealEstate10KPSNR (dB): 24.97 -> 26.25Improved
add REPA distillation (ablation)Efficient-LVSM 6+6 · RealEstate10KPSNR (dB): 26.02 -> 26.81Improved
add REPA distillation (ablation)LVSM Dec-Only · RealEstate10KPSNR (dB): 25.52 -> 25.68Improved
use Smooth L1 REPA loss instead of L2 or cosine (ablation)Efficient-LVSM REPA · RealEstate10KPSNR (dB): 26.39 -> 26.81 (L2); 26.30 -> 26.81 (Cosine)Improved
distill both input and target tokens instead of only one (ablation)Efficient-LVSM REPA · RealEstate10KPSNR (dB): 26.35 -> 26.60Improved
use DINOv3 layer 8 instead of layer 10 or 12 (ablation)Efficient-LVSM REPA · RealEstate10KPSNR (dB): 26.28 -> 26.60 (layer 10); 26.11 -> 26.60 (layer 12)Improved
increase model from 3+3 to 6+6 to 12+12 layers (ablation)Efficient-LVSM · RealEstate10KPSNR (dB): 26.43 -> 27.77 -> 28.32Improved
add attention masks to LVSM (ablation)LVSM · RealEstate10KPSNR (dB): 25.24 -> 24.13Hurt
add MMDiT-style separate projections to LVSM (ablation)LVSM · RealEstate10KPSNR (dB): 25.24 -> 24.37Hurt
use decoupled dual-stream design instead of LVSM Dec-OnlyEfficient-LVSM · RealEstate10K inference latencylatency (ms): 109.37 ms -> 24.78 msImproved
use decoupled dual-stream design instead of LVSM Dec-Only (512 res)Efficient-LVSM · RealEstate10K (res-512)PSNR (dB): 29.53 -> 29.86Improved

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