AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Beyond Time Shifts: Adapting Omni-LLM as a Reference-Free Evaluator for Generative Audio-Visual Models

Yijie Qian, Juncheng Wang, Chao Xu, Huihan Wang, Yuxiang Feng, Yang Liu, Baigui Sun, Yong Liu, Shujun Wang

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Interleaved image-text generation with MLLMs · 1 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
replace regression with scalar ranking (BTL) head (ablation)Omni-LLM evaluator · SynthSync testKendall tau: 0.1628 -> 0.4515Improved
score as generated text instead of continuous scalar head (ablation)Omni-LLM evaluator · SynthSync testPair acc: 71.07 -> 56.19Hurt
train on SynthSync instead of zero-shot Qwen2.5-Omni-3B (ablation)Omni-LLM evaluator · SynthSync testPair acc: 51.02 -> 66.60Improved
add preference alignment fine-tuning (PFT) (ablation)Omni-LLM evaluator · SynthSync testPair acc: 66.60 -> 71.16Improved
add R-GRPO reinforcement post-training (ablation)Omni-LLM evaluator · SynthSync testPair acc: 71.16 -> 72.38Improved
use listwise reward only in R-GRPO (ablation)R-GRPO · SynthSync testKendall tau: 0.4515 -> 0.4267Hurt
use pairwise reward only in R-GRPO (ablation)R-GRPO · SynthSync testKendall tau: 0.4515 -> 0.4700Improved
combine pairwise and listwise rewards (dual) (ablation)R-GRPO · SynthSync testKendall tau: 0.4515 -> 0.4899Improved
set exploration sigma to 10.0 instead of 2.5 (ablation)R-GRPO · SynthSync testPair acc: 72.38 -> 71.70Hurt
set exploration sigma to 2.0 instead of 2.5 (ablation)R-GRPO · SynthSync testPair acc: 72.38 -> 71.56Hurt
use the learned metric as Best-of-N reward instead of AV-AlignLTX-2 BoN · LTX-2 BoN, AV-Align scorescore: 0.0698 -> 0.1439 (Random vs +Ours)Improved
use VQ-Insight instead of our evaluator for bottom-1 detectionOmni-LLM evaluator · SynthSync testBottom-1 acc: VQ-Insight 62.80 vs Ours 55.00Mixed

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