AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

PRIMT: Preference-based Reinforcement Learning with Multimodal Feedback and Trajectory Synthesis from Foundation Models

Ruiqi Wang, Dezhong Zhao, Ziqin Yuan, Tianyu Shao, Guohua Chen, Dominic Kao, Sungeun Hong, Byung-Cheol Min

NeurIPS 2025 · Vision-language and multimodal › Multimodal LLMs · approach: Reinforcement learning for visual reasoning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

placeholder

Problem

placeholder

Approach

placeholder

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use PRIMT's multimodal feedback fusion and bidirectional trajectory synthesis instead of single-modality FM feedbackPbRL on the evaluated robotics tasks · 8 DMC, MetaWorld, and ManiSkill taskstask performanceImproved
use PRIMT instead of the scripted-teacher PrefGT baselinePbRL · Sweep Into and PegInsertionSidetask performanceImproved
use PRIMT's hierarchical neuro-symbolic fusion instead of naive multimodal evaluationsynthetic preference feedback · Evaluated taskspreference-label accuracyImproved
remove the intra-modal crowd-check and preference-fusion module (ablation)PRIMT · Door Open and PickSingleYCBtask performanceHurt
remove PSL-based inter-modal preference fusion (ablation)PRIMT · Door Open and PickSingleYCBtask performanceHurt
remove foresight trajectory generation (ablation)PRIMT · Door Open and PickSingleYCBearly-stage learning speedHurt
add foresight trajectory generation to initialize the trajectory buffer (ablation)PRIMT · Evaluated taskspreference indecision rateImproved
remove hindsight trajectory augmentation (ablation)PRIMT · Door Open and PickSingleYCBfinal task performanceHurt
remove the causal auxiliary loss while retaining counterfactual trajectory augmentation (ablation)PRIMT · Door Open and PickSingleYCBfinal task performanceHurt
replace hierarchical multimodal fusion with direct multimodal FM evaluationPbRL · Evaluated taskstask performanceHurt
replace GPT-4o with GPT-4o-mini (ablation)PRIMT · Evaluated tasksperformance: 16–26%Hurt

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