With the rapid development and adoption of 3D Gaussian Splatting (3DGS), the need for effective copyright protection has become increasingly critical. Existing watermarking techniques for 3DGS mainly focus on protecting rendered images via pre-trained decoders, leaving the underlying 3D Gaussian primitives vulnerable t...
Hao Qin, Yukai Sun, Lu-Yuan Chen et al.· 0 citations
Reinforcement Fine-Tuning (RFT) has enabled medical Multimodal Large Language Models (MLLMs) to produce Chain-of-Thought (CoT) reasoning for visual question answering, yet these models suffer from $\textit{confidence miscalibration}$---a systematic gap between expressed certainty and actual diagnostic accuracy that und...
Yuetian Du, Yu-Cheng Wang, Zhenyuan Chen et al.· Lecture notes in computer sc...· 0 citations
Reinforcement learning for multi-turn search reasoning typically relies on terminal outcome rewards, which cannot distinguish useful, redundant, and harmful intermediate interactions. We propose LOTAPO , a self-generated process-supervision method based on backward leave-one-turn attribution. For each search turn, LOTA...
LAPOlations show complementary benefits from backward attribution and sign-consistency gating, demonstrating that policy-derived retrospective attribution can provide effective process supervision for multi-turn search agents.
MetaRAG is proposed, a belief-action aligned policy optimization framework for agentic RAG that consistently improves the accuracy-efficiency trade-off over strong RL-based agentic RAG baselines, with gains that transfer to deep research settings, different optimizers, and multiple model backbones.
Qiuyi Qi, Tian Liang, Jiamu Wang et al.· 0 citations
A diagnostic benchmark for open-domain, open-form, long-horizon counterfactual causal reasoning, containing 220 what-if questions across STEM, HSS, and Hybrid scenarios, and finds that WhatIfBench remains far from saturated: even the strongest model reaches only a 64.62% final score.
Yu-Cheng Wang, Yuetian Du, Zheng Liu et al.· 0 citations
Living-Harness is proposed, a self-evolving agent harness that converts each completed trajectory and its evaluator signals into posterior evidence for bounded harness updates, and supports retrieval-only reuse of the evolved harness state across model backbones.
Yuetian Du, Yucheng Wang, Helsing Xu et al.· arXiv.org· 2 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
