作者位次 #5/8ICLR 2026Poster· Reinforcement Learning / Deep RL
Don't Just Fine-tune the Agent, Tune the Environment
Siyuan Lu · Zechuan Wang · Hongxuan Zhang · Qintong Wu · Leilei Gan · Chenyi Zhuang · Jinjie Gu · Tao Lin
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Siyuan Lu · Zechuan Wang · Hongxuan Zhang · Qintong Wu · Leilei Gan · Chenyi Zhuang · Jinjie Gu · Tao Lin
Ziwei Huang · Ying Shu · Fanghao · Quanyu Long · Wenya Wang · Qiushi Guo · Tiezheng Ge · Leilei Gan
Subject-driven image generation models face a fundamental trade-off between identity preservation (fidelity) and prompt adherence (editability). While online reinforcement learning (RL), specifically GPRO, offers a promising solution, we find that a naive application of GRPO leads to competitive...
Wenyi Xiao · Xinchi XU · Leilei Gan
Large Vision Language Models (LVLMs) achieve strong multimodal reasoning but frequently exhibit hallucinations and incorrect responses with high certainty, which hinders their usage in high-stakes domains. Existing verbalized confidence calibration methods, largely developed for text-only LLMs, t...
Yandi Wang · Libin Zhan · Ziwei Huang · Tiancheng Luo · Yuxuan Jiang · Wang Dong · Leilei Gan · Jun Chen
Extracting structured information from visual documents (Visual Information Extraction, VIE) is a cornerstone of business automation. While recent Multimodal Large Language Models (MLLMs) have shown promising capabilities, existing benchmarks suffer from critical limitations in scale and realism,...
FuLin Shi · Wenyi Xiao · Leilei Gan · Liang Ding · Binchen
Evaluating the alignment between textual prompts and generated images is critical for ensuring the reliability and usability of text-to-image (T2I) models. However, most existing evaluation methods rely on coarse-grained metrics or static Question Answering (QA) pipelines, which lack fine-grained...