AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

Test-Time Computing for Referring Multimodal Large Language Models

Mingrui Wu, Hao Chen, Jiayi Ji, Xiaoshuai Sun, Zhiyuan Liu, Liujuan Cao, Ming-Ming Cheng, Rongrong Ji

IJCV 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Test-time adaptation of VLMs · 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
add Optim++ Adam optimizer in place of SGD (ablation)ControlMLLM++ · ROC (box, test)accuracy: 60.59 -> 66.93Improved
add answer-start token attention A(st) on top of Adam (ablation)ControlMLLM++ · ROC (box, test)accuracy: 66.93 -> 65.70Hurt
add layer selection on top of Adam (ablation)ControlMLLM++ · ROC (box, test)accuracy: 66.93 -> 65.70Hurt
add both A(st) and layer selection on top of Adam (ablation)ControlMLLM++ · ROC (box, test)accuracy: 66.93 -> 68.02Improved
add PromptDebias contrastive decoding (ablation)ControlMLLM++ · ROC (box, test)accuracy: 68.02 -> 71.19Improved
add early stopping to ControlMLLM (T=0 vs T=5) (ablation)ControlMLLM · ROC (val)accuracy: 57.00 -> 62.50Improved
lower EMA decay beta to 0.3 at alpha=400 (ablation)ControlMLLM · ROC (val)accuracy: 53.5 -> 62.5Improved
add ControlMLLM test-time optimization to LLaVA-1.5LLaVA-1.5 · ROC (box, test)accuracy: 54.72 -> 60.59Improved
use ControlMLLM++ instead of ControlMLLM on Qwen2.5-VL-7BQwen2.5-VL-7B · ROC (box, test)accuracy: 79.20 -> 79.20No clear effect
add ControlMLLM++ to Qwen2.5-VLQwen2.5-VL · RTC (box, test)accuracy: 81.91 -> 88.23Improved
scale LLaVA-1.5 from 7B to 13B (vanilla)LLaVA-1.5 · ROC (box, test)accuracy: 54.72 -> 55.69No clear effect
add ControlMLLM++ to LLaVA-HRLLaVA-HR · Screenshot referring descriptionCIDEr-D: 10.88 -> 66.96Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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