DCGC is a Masked Diffusion Model framework that performs global correction of flawed reasoning traces in Large Language Models. It uses an imperfect solution draft from an upstream solver as auxiliary context and combines task‑specific supervised fine‑tuning with a Dynamic Dual‑CFG inference mechanism that separates problem‑only and joint problem‑draft branches. Experiments on math, code, and knowledge reasoning benchmarks show that DCGC outperforms standard sampling and simpler CFG variants, and it can improve full test‑set accuracy even when ground‑truth failure labels are unavailable.
By Minhae Oh, Nakyung Lee, Jungwoo Lee
Traffic Anomaly Understanding (TAU) requires models and systems to detect, reason about, and explain anomalous events in transportation videos. To address this challenge, we propose TAU-Agent, an agen...
Automated fact-checking (AFC) systems retrieve evidence and predict claim veracity, yet evaluations omit simple baselines, systems are developed for a single benchmark and cannot be trusted to general...
A recent line of synthetic-data work reconstructs the thinking behind existing text rather than rewriting the text itself, but it operates on short web passages, recovers only local thoughts, and leav...
Acquiring pixel-level annotations for medical image segmentation is a severe bottleneck. Traditional U-Net architectures, while effective, learn local texture patterns and lack awareness of global ana...
Large language models (LLMs) demonstrate impressive performance across a wide range of general NLP tasks; however, their effectiveness in sensitive domains, such as hate speech detection, remains less...
Multi-camera systems are foundational to modern media production, and multi-camera editing is a critical task. This involves the proper selection of the appropriate camera view at each moment. In this...
To reduce deployment cost and retraining overhead, adapting pretrained learned image compression (LIC) models to downstream machine vision tasks has attracted growing attention. However, existing meth...
We introduce VGA-BenchV2, an extended human-aligned benchmark and optimization framework for jointly evaluating and improving video generation quality and aesthetic value. Built upon VGA-Bench, VGA-Be...
arXiv:2608.23664v1 Announce Type: cross
Abstract: Reward fine-tuning is becoming an important tool for adapting diffusion models to human preferences and task-specific objectives, but existing method...
By Jaemoo Choi, Wei Guo, Yuchen Zhu, Arash Vahdat, Molei Tao, Julius Berner, Yongxin Chen
arXiv:2603.16654v3 Announce Type: replace-cross
Abstract: Evaluating the reasoning abilities of large language models (LLMs) solely from final answers can obscure failures in intermediate steps, espe...
By Xiaojie Gu, Sherry T. Tong, Aosong Feng, Sophia Simeng Han, Jinghui Lu, Yingjian Chen, Yusuke Iwasawa, Yutaka Matsuo, Chanjun Park, Rex Ying, Irene Li
The paper introduces CAIT, a benchmark of 400 synthetic scenes featuring counter‑intuitive actions that challenge multimodal large language models (MLLMs). Human participants and proprietary models like Claude and Gemini perform well, but standard open‑source instruction‑tuned MLLMs fail, largely due to a strong language prior that overrides contradictory visual evidence. The study shows that Chain‑of‑Thought reasoning can help but introduces new issues, while targeted fine‑tuning and structured prompting can reduce reliance on language priors and improve visual grounding.
By Chen Ling, Tongwei Zhang, Hanqian Li, Nai Ding
The paper introduces ARLI, a latency‑aware framework that enables reinforcement learning fine‑tuning of large generalist robot policies despite inference delays. ARLI combines asynchronous inference with state augmentations—incorporating committed actions and mid‑inference observations—to restore near‑Markovian dynamics and maintain reactivity. Experiments on simulated and real‑world manipulation tasks show that ARLI allows effective policy improvement under latency, outperforming standard RL even in no‑latency scenarios.
By Brian Zhu (Siemens), Momen Khalil (Siemens), E Harrison (UC Berkeley), Emanuele Poggi (Siemens), Philipp Schmitt (Siemens), Bernd Kast (Siemens), Philine Meister (Siemens), Pranav Atreya (UC Berkeley), Qiyang Li (UC Berkeley), Finn Ferchau (Siemens), Cesar Colmenero (Siemens), Yash Shahapurkar (Siemens), Gokul Narayanan (Siemens), Melih Erdogan (Siemens), Kai Wurm (Siemens), Georg von Wichert (Siemens), Oier Mees (Microsoft, ETH Zurich, UC Berkeley), Eugen Solowjow (Siemens), Andrew Wagenmaker (UC Berkeley), Sergey Levine (UC Berkeley)
NeuronGuard is a fine‑tuning defense for large language models that hardens them against both jailbreak and neuron‑level attacks. It redistributes safety signals across many neurons by identifying safety‑critical ones with per‑layer linear classifiers, enforcing refusal behavior when those neurons are ablated, and applying KL‑divergence regularization for consistency. A randomized gradient projection preserves task performance, and the authors provide a formal guarantee that NeuronGuard lowers the attack success rate upper bound, with experiments showing near‑zero success rates across multiple models and attack strategies.
By Anjun Gao, Yueyang Quan, Yufei Xia, Zhuqing Liu, Minghong Fang
The paper introduces Influence‑Guided PPO (I‑PPO), a framework that incorporates data attribution into the Proximal Policy Optimization (PPO) post‑training loop for large language models. I‑PPO computes an influence score for each episode via a gradient‑based approximation and removes episodes that are anti‑aligned with a validation gradient, thereby filtering out noisy or unfaithful reasoning. Experiments show that this filtering consistently outperforms both supervised fine‑tuning (SFT) and standard PPO, acting as an intrinsic early‑stopping mechanism that speeds up training and reduces unfaithful chain‑of‑thought reasoning.
By Dong Shu, Denghui Zhang, Jessica Hullman
The paper introduces a two‑layer detector to prevent ‘slopsquatting’—the risk of local coding LLMs fabricating Python package names that adversaries can pre‑register on PyPI. The first layer checks PyPI for existence, while the second uses a Random Forest classifier on ten name‑and‑metadata features; an import reconciler resolves naming mismatches. Embedded in a LangGraph state machine, the system retries at escalating temperatures and falls back to stronger models, achieving hallucination‑free code in 76% of 300 curated prompts and recovering additional runs through intra‑ and cross‑model retries.
By Akash Raj, Sargam Sahu
The paper introduces Hierarchical Skill Retrieval (HSR), a framework that decomposes a target manipulation task into candidate skill sequences and evaluates each plan for semantic plausibility and skill reliability. HSR combines subtask-level language retrieval with behavior-feature reranking to select demonstrations that are both relevant and compatible with the target task, followed by a two-stage pretraining and finetuning pipeline for policy adaptation. Experiments on the LIBERO benchmark and real-world robot tasks show that HSR improves average success rates by 10.3% and 21.3% over the strongest baseline, demonstrating the effectiveness of structured skill-level retrieval for data-efficient Vision‑Language‑Action adaptation.
By Haoran Hao, Shahram Najam Syed, Jeff Schneider, Jeffrey Ichnowski
arXiv:2608.23830v1 Announce Type: cross
Abstract: RL has emerged as a powerful paradigm for enhancing the instruction following capabilities of LLMs. While existing training recipes achieve substanti...
By Mian Zhang, Yueqin Yin, Kaiyu He, Peilin Wu, Xinlu Zhang, Mingyuan Zhou, Zhiyu Zoey Chen
The paper introduces UCO, a multi‑turn interactive reinforcement learning method designed to improve adaptive teaching with large language models. UCO employs two reward functions—Progress Reward to gauge genuine cognitive advancement and Scaffold Reward to keep instruction within each student’s Zone of Proximal Development. Experiments on BigMath and MathTutorBench show UCO outperforming 11 baseline models and matching advanced closed‑source systems.
By Shouang Wei, Min Zhang, Xin Lin, Bo Jiang, Kun Kuang, Zhongxiang Dai
arXiv:2608.24093v1 Announce Type: cross
Abstract: Self-supervised representation learning for 4D point cloud videos is challenging because annotations are costly and reconstruction-based pretraining...
By Jheng-Ling Lee, Shang-Tse Chen