AI Paper Atlas

Research map / Video / Video generation

Aligning Human Sense: Calibrated Distributional Reward Learning for Video Generation

Naixin Zhai, Weihua Cheng, Dexu Yu, Yikai Gu, Hanwen Du, Junchen Fu, Chenxi Huang, Yingwei Song, Liyuan Ma, Yang Ran, Youhua Li, Yongxin Ni

ECCV 2026 · Video › Video generation · approach: Compositional text-to-video diffusion · 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
train reward model on calibrated full data instead of raw (ablation)reward model · Reward model, Motion Effects (w/ ties)pairwise accuracy: 65.66 -> 71.72Improved
train reward model on calibrated full data instead of raw (ablation)reward model · Reward model, Action (w/o ties)pairwise accuracy: 80.30 -> 83.33Improved
curate equal-scale data instead of raw (ablation)reward model · Reward model, Action (w/ ties)pairwise accuracy: 54.65 -> 59.60Improved
add Wasserstein loss to reward model (ablation)reward model · Reward model, Motion Effects (w/ ties)pairwise accuracy: 68.33 -> 71.72Improved
add Wasserstein loss to reward model (ablation)reward model · Reward model, Action (w/ ties)pairwise accuracy: 58.98 -> 60.61Improved
replace KL with Wasserstein in GRPO (ablation)WGRPO · Video-Bench generationavg score: 68.02 -> 69.58Improved
add KL regularizer to GRPO (ablation)GRPO · Video-Bench generationavg score: 67.97 -> 68.02No clear effect
KL instead of Wasserstein on dynamic degree (ablation)GRPO · Video-Bench generationdynamic degree: 36.50 vs 36.48Mixed
calibrated reward model for video generationWan2.1-T2V-1.3B · Video-Bench generationavg score: 68.16 -> 69.58Improved
calibrated reward model for video generationCogVideoX-2B · Video-Bench generationsubject consistency: 87.35 -> 86.45Hurt
calibrated reward model instead of rawDanceGRPO · Video-Bench generation (optimizer comparison)avg score: 68.99 -> 69.97Improved
calibrated reward model instead of rawFlowGRPO · Video-Bench generation (optimizer comparison)avg score: 69.86 -> 70.87Improved

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