Academic paper search is a core step in scientific research, and LLM-based search agents are emerging as a promising paradigm for iterative, intent-driven literature exploration. However, existing benchmarks offer limited coverage of research intents and insufficient support for scalable answer construction and reprodu...
Tingyue Pan, Mingyue Cheng, Daoyu Wang et al.· 0 citations
Self-play reinforcement learning has recently emerged as a way to train driving policies without any human data. It uses cheap, large-scale simulations to substitute expensive, large-scale human driving demonstrations. A key limitation of this approach is that policies trained through pure self-play can learn effective...
Daphne Cornelisse, Julian Hunt, Zixu Zhang et al.· 0 citations
Model merging enables training-free multi-task learning by aggregating knowledge from models fine-tuned from a shared checkpoint, but their task updates can interfere. Using proxy data, we find that task updates induce output shifts whose energy concentrates in a few dominant directions, while many low-energy direction...
Long-context reasoning is an essential capability for large language models, particularly when they are deployed as autonomous agents that must reason over lengthy trajectories. Reinforcement learning (RL) has recently emerged as a dominant paradigm for improving this ability, yet existing work largely focuses on rewar...
Xiaoyue Xu, Sikui Zhang, Xiaorong Wang et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Continual post-training enables models to absorb emerging knowledge after deployment, but repeatedly updating shared parameters can accumulate weight drift, potentially causing catastrophic forgetting and degrading general capabilities. Retrieval-augmented generation avoids such parameter drift, yet often lacks the dep...
Weihang Su, Jiacheng Kang, Jingyan Xu et al.· 0 citations
Passing an exact-clone null gate does not entail passing a matched target-selection utility gate. We call a non-tie preference between exact duplicates dark current. On 102 prespecified fresh RewardBench 2 rows, all four configuration--criterion cells report 0/102 clone preferences. Both neutral utility arms pass, but...
Hiroyasu Usami, Keisuke Hara, Ayato Tsuboi et al.· 0 citations
LLM agents are increasingly used to simulate real-world interactions, but it remains unclear whether simulated behaviors preserve the content patterns and interaction dynamics of real human behaviors. Existing evaluations often focus on individual comments or task performance, rather than whether simulated behaviors ma...
Yaoning Yu, Ye Yu, John Sciarappo et al.· 0 citations
Vision-language models (VLMs) are increasingly used for scene understanding in autonomous driving, but robustness analysis often relies on task-agnostic embedding stability alone. We study whether corruption-induced embedding drift predicts changes in a task-aligned hazard score derived from CLIP image-text similaritie...
RAIL is introduced, a human-centric benchmark grounded in the Cattell--Horn--Carroll (CHC) framework, which formalises auditory cognition into five core capabilities, and develops them into structured evaluation tasks with principled data curation and human-aligned evaluation protocols.
Hong-Yu Jin, Siyi Wang, Yang Xiao et al.· arXiv.org· 3 citations
On-device inference reduces exposure to remote infrastructure, but it does not establish privacy for an operating-system assistant that aggregates cross-application data, retains derived state, or delegates actions. We present a feature-boundary audit method for OS-integrated AI. It combines a taxonomy of six privacy-f...
Large language model agents use skills that package reusable instructions, code, and resources. Existing skill-injection attacks exploit trust in these packages by adding malicious content. As agents become more capable, however, they can skip unnecessary steps, inspect code, and reject unrelated actions. We introduce...
Haochang Hao, Dehai Min, Zhifang Zhang et al.· 0 citations
High-quality training data is essential to large language models (LLMs) and typically requires extensive and costly manual curation. Automatic data preparation systems support reusable operators and task-driven workflows, yet adapting these workflows to the desired data quality remains challenging. High-quality example...
Chao Deng, Shaolei Zhang, Ju Fan et al.· 0 citations
