When retrieved evidence contradicts an agent's prior beliefs, does it revise its answer, acknowledge uncertainty, or persist with an incorrect conclusion? Existing evaluations of agentic systems focus primarily on task success, offering limited insight into how agents handle such conflicts. We propose to evaluate agent...
Kaiser Sun, Bernal Jiménez Gutiérrez, Hong-Jun Liu et al.· 0 citations
Social media systems typically display conversations among already familiar contributors, which can be predictable and one-sided. In civic discourse, this design narrows discussion, reinforces divides, and distorts the perception of public opinion. To encourage cross-community engagement, we present CrossWeave, an AI-p...
Fei Fang, Reva Hirave, William Jurayj et al.· Proceedings of the 2026 Comp...· 0 citations
SIGMA is proposed, a data generation and training pipeline enabling alignment self-improvement that generalizes to out-of-distribution settings and shows that a Model Spec balancing harmlessness and helpfulness, test-time reasoning for safety deliberation, and high-quality rubrics from SIGMA's task designer agent are c...
Jing-Yu Zhang, Shruti Palaskar, Daniel Khashabi et al.· 0 citations
World models enable agents to plan by predicting future states of the environment, but their predictions can become unreliable when test-time dynamics differ from those seen during training. We present JEPA-TTT, which adapts the latent dynamics predictor of a pretrained action-conditioned Joint-Embedding Predictive Arc...
Zheyuan Zhang, Su-Yu Ye, Nakul Agarwal et al.· 0 citations
A language-model agent is jointly defined by its model and its harness, the executable program that organizes model calls, tool use, and information flow. Because different tasks call for different ways of organizing these operations, the harness needs to be adapted using feedback from the task at hand. We introduce ha...
Alvin Zhang, Xue-Chen Liu, Zi-Xuan Wang et al.· 0 citations
The Very Big Video Reasoning (VBVR) Dataset is introduced, an unprecedentedly large-scale resource spanning 200 curated reasoning tasks following a principled taxonomy and over one million video clips, approximately three orders of magnitude larger than existing datasets.
Maijunxian Wang, Rui-Si Wang, Ju-Yi Lin et al.· arXiv.org· 28 citations· ⚡5
It is demonstrated that enforcing fine-grained (sentence-level) citations forfeits gains of 2-97% relative to the best-performing granularity, and up to 338% on individual tasks, and that effective attribution requires matching granularity to the model's semantic scope rather than fixing it by convention.
Hexuan Wang, Jing-Yu Zhang, Benjamin Van Durme et al.· arXiv.org· 2 citations· ⚡1
A controlled cross-modality framework that instantiates the same task suite in a variety of modalities to test the Convergent Emergence Hypothesis and shows that paired-mapping ICL emerges across six modalities, surpasses controlled baselines, and has correlated per-task effects across five of them.
Nathan Breslow, Seungwook Han, D. Lee et al.· 0 citations
Together, these results show that composing complementary mechanisms substantially improves long-horizon memorization beyond what any individual mechanism achieves.
Zhe-Yuan Zhang, Alvin Zhang, Daniel Khashabi et al.· 1 citation
This work presents AutoMat, a benchmark for evaluating LLM-based agents'ability to reproduce claims from computational materials science, and curates a set of claims from real materials science papers to test whether coding agents can recover and execute the end-to-end workflow needed to support (or undermine) such cla...
Ziyang Huang, Yiyun Cao, A. Shargh et al.· arXiv.org· 9 citations
This work introduces VBVR-Pro, a closed-loop testbed that makes native visual reasoning through generation trainable, verifiable, optimizable, and experimentally controllable, and identifies recurring failure modes of the prevalent VLM-as-a-judge paradigm.
Junhua Xu, Rui-Si Wang, Fanyi Pu et al.· 5 citations
The Information Abundance Paradox is supported and it is suggested that scaling toward near-infinite context is not simply a matter of supplying more data, even when high-quality long-context data is abundant.
Arda Uzunouglu, Benjamin Van Durme, Daniel Khashabi· 2 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
