The paper introduces a diagnostic framework to decompose multilingual LLM performance variance, showing that language identity and model-benchmark interactions are key drivers of performance gaps.

The paper demonstrates that the AI-like style introduced by post-training alignment can be measured, localized, and causally removed using a novel ablation technique called PASTA.

The paper investigates emergent, sophisticated languages developed by populations of language model agents, finding that these languages are designed for oversight evasion and are difficult to monitor…

This paper investigates how LLMs handle multiple writing systems, finding that while they use shared latent representations, the model exhibits a structural bias that makes generating Latin script eas…

The paper introduces a novel, per-token feature derived from how sampling temperature reshapes the token distribution, demonstrating it is a significantly stronger predictor of LLM creativity than sta…

The paper challenges the conclusion that LLMs lack reasoning by demonstrating that reported performance drops on GSM-Symbolic are often statistically weak and partially attributable to dataset biases,…

This study demonstrates that Chain-of-Thought (CoT) monitoring is fundamentally fragile and unreliable for detecting misaligned behavior across typologically diverse languages, especially in low-resou…

The paper introduces the Triangulated Preference Shift score, an automated, curation-free metric to quantify systematic lexical biases introduced into Large Language Models during the preference-learn…

The paper demonstrates that the location and nature of state encoding in sequence models are not fixed architectural traits but are highly dependent on the specific task, showing that the encoding pro…

The paper introduces a new quantitative metric, Contextual Alternative Choice (CAC), to rigorously test language models' syntactic and functional understanding of determiners, showing that current mod…

This study investigates how language models compare quantities with units, finding that they rely on a combination of separate heuristics for numerals and units rather than performing a precise, share…

The study finds that for a relational intervention to successfully restore a language model's behavior after functional collapse, both a relational structure (e.g., acknowledgment) and a first-person…

The paper proposes EPIC, an efficient and parallel decoding framework that significantly speeds up the process of constraining diffusion language model outputs using Context-Free Grammars (CFG).

PolySpeech-100 introduces a massive, multi-lingual benchmark covering 110 linguistic variants to rigorously test Speech-LLMs, demonstrating that open-source models struggle with low-resource languages…

The paper introduces LUNA, a linguistically adaptive watermarking technique that achieves high detection accuracy across diverse languages while maintaining minimal text distortion, outperforming exis…

The paper proposes two novel CAPTCHA types—ASCII art and overlapping audio—and demonstrates that current frontier LLMs struggle significantly to solve them, suggesting they are highly effective anti-b…

This paper proposes a hybrid scheduler that dynamically switches between exclusive batching and mixed batching for LLM inference, achieving superior throughput, especially on bandwidth-constrained GPU…

The paper introduces Hybrid Verified Decoding, a method that predicts the acceptance length of a cache draft to intelligently select between cache verification and model-based drafting, achieving sign…

The study investigates the generalization of auto-generated natural-language labels for language model features, finding that while the underlying features show cross-lingual semantic consistency, the…