AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Structure-based RNA Design by Step-wise Optimization of Latent Diffusion Model

Qi Si, Xuyang Liu, Penglei Wang, Xin Guo, Yuan Qi, Yuan Cheng

AAAI 2026 · Image generation › Autoregressive and latent generation · approach: Diffusion for protein and antibody design · 3 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 step-wise RL fine-tuning to the LDMLDM · SOLD TESTMFE: -13.3275 -> -16.8611Improved
add step-wise RL fine-tuning to the LDMLDM · CASP15 TESTSS: 0.5543 -> 0.6957Improved
add step-wise RL fine-tuning to the LDMLDM · SOLD TESTsequence recovery: 0.5728 -> 0.5732No clear effect
add step-wise RL fine-tuning to the LDMLDM · SOLD TESTSS: 0.7269 vs 0.7601 (but RiboDiffusion 0.7632)Hurt
use RNA-FM latent space instead of one-hot sequence diffusionRiboDiffusion-style LDM · SOLD TESTsequence recovery: 0.5125 -> 0.5728Improved
use single-step sampling instead of full trajectory RLSOLD vs DDPO/DPOK · MFE trainingtraining time per epoch (s): 5953 -> 256 (DDPO)Improved
use single-step sampling instead of full trajectory RLSOLD vs DDPO/DPOK · LDDT trainingtraining time per epoch (s): 14000 -> 6900 (DDPO)Improved
SOLD vs DDPO on SS rewardSOLD · SOLD TESTSS: 0.7595 (DDPO) vs 0.7551 (SOLD)No clear effect
SOLD vs DDPO on SS rewardSOLD · CASP15 TESTSS: 0.6649 -> 0.7010Improved
RL on MFE reward vs DRAKESSOLD · SOLD TESTMFE: -14.2374 -> -19.7428Improved

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