Across three models, two reasoning regimes, and two domains, four findings hold without per-turn reasoning: displaying accurate state is unreliable, an unverified ledger the agent writes itself beats an accurate checklist it is shown, directives help in proportion to the model's obedience, and enforcement needs no obed...
C. Zhang, Wonbin Kweon, Jia-Wei Han· arXiv.org· 0 citations
Orthrus is a serving system that performs heterogeneous batching within a unified inference loop that addresses the primary challenge lies in unifying embedding and generation workloads with conflicting computational patterns while optimizing batch composition for high performance.
Dohyun Park, Hubertus Franke, Daniel G. Waddington et al.· Proceedings of the 2026 Conf...· 1 citation
For predicting aggregate engagement, a plain LLM ranker beats persona simulation - synthetic personas are not merely a weak predictor, they are worse than not using them.
It is shown that preference alignment preserves the human response distribution only under a restrictive condition, and no consistent evidence that real human preferences satisfy it, and human-likeness is established as an explicit dimension of alignment rather than something assumed to follow from preference alignment...
Su-Qin Yuan, Runqi Lin, Mu-Yang Li et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Retrieval-Augmented Generation (RAG) aims to enhance the trustworthiness of Large Language Models (LLMs) by grounding their outputs in external documents, often using inline citations for verifiability. However, the faithfulness of these citations -- whether the model genuinely uses a source to generate an answer -- re...
Ian van Dort (University of Amsterdam), Maria Heuss (University of Amsterdam)· 0 citations
The adversarial robustness of three frontier large language models developed by Anthropic, Opus 4.8, Fable 5 and its successor Fable 5.1 is evaluated against four families of automated jailbreak attack across 7,826 harmful intents spanning a ten-category harm taxonomy.
Building a lexicon from discovered word-like units is a central goal in zero-resource speech processing. But do our evaluations provide a trustworthy indication of lexicon quality? A common metric, normalized edit distance, averages the phoneme edit distances between discovered units in each cluster. We show that this...
Simon Malan, Danel Slabbert, Herman Kamper· 0 citations
Ro is proposed, a rubric-guided process reward framework for stepwise model routing that consistently outperforms strong baselines and achieves better accuracy and cost trade-offs.
Sheng-Hao Ye, Yu Guo, Zheng-Heng Li et al.· arXiv.org· 2 citations
In this work, we propose BAIT (Boundary-Aware Iterative Trap), a three-step jailbreak framework that elicits malicious information through internal disclosure by target large language models (LLMs), instead of external feedback from judge LLMs. BAIT first asks the model to identify the protection boundary, then require...
3D Gaussian Splatting (3DGS) has emerged as a powerful technique for novel view synthesis, subsequently extending into numerous spatial AI applications. However, most existing 3DGS methods operate in isolation, focusing on specific domains. In this paper, we introduce X-GS, an extensible framework that integrates previ...
Althea is introduced, a retrieval-augmented system for user-driven claim evaluation that matches standard pipelines on AVeriTeC while improving supported/refuted discrimination and cautioning against treating AI-delivered verdicts as a source of durable literacy gains.
S. Churina, Kokil Jaidka, A. Barik et al.· 0 citations
Formal semantics and distributional semantics are distinct approaches to linguistic meaning: the former models meaning as reference via model-theoretic structures; the latter as vectors in high-dimensional spaces shaped by usage. This paper establishes which part of intensional formal semantics admits a linear vector-s...
Flint-interactive gives AI agents a language for creating interactive visualizations. Click a chart, filter data, highlight related content, and explore beyond the view the author originally chose. The post Flint-interactive: A language for creating interactive data visualizations appeared first on Microsoft Research.
