Related papers: When Does a Language Model Commit? A Finite-Answer…
We study the source of uncertainty in DeepSeek R1-32B by analyzing its self-reported verbal confidence on question answering (QA) tasks. In the default answer-then-confidence setting, the model is regularly over-confident, whereas semantic…
Do large language models (LLMs) anticipate when they will answer correctly? To study this, we extract activations after a question is read but before any tokens are generated, and train linear probes to predict whether the model's…
Transformer-based language models (LMs) have achieved widespread empirical success, but their theoretical expressive power remains only partially understood. In this work, we analyze a restricted idealization of fixed-precision transformers…
Large language models have achieved significant reasoning improvements through reinforcement learning with verifiable rewards (RLVR). Yet as model capabilities grow, constructing high-quality reward signals becomes increasingly difficult,…
Vision-language models (VLMs) are increasingly used to answer questions about physical scenes, yet most evaluations reduce performance to a final answer. This hides whether the model perceived the right objects, represented the right…
Extensively evaluating the capabilities of (large) language models is difficult. Rapid development of state-of-the-art models induce benchmark saturation, while creating more challenging datasets is labor-intensive. Inspired by the recent…
Positional bias in binary question answering occurs when a model systematically favors one choice over another based solely on the ordering of presented options. In this study, we quantify and analyze positional bias across five large…
This paper investigates the knowledge of language models from the perspective of Bayesian epistemology. We explore how language models adjust their confidence and responses when presented with evidence with varying levels of informativeness…
We show that continual pretraining on plausible misinformation can overwrite specific factual knowledge in large language models without degrading overall performance. Unlike prior poisoning work under static pretraining, we study repeated…
Large language models (LLMs) must often respond to highly ambiguous user requests. In such cases, the LLM's best response may be to ask a clarifying question to elicit more information. Existing LLMs often respond by presupposing a single…
Open-domain question answering (QA) is known to involve several underlying knowledge and reasoning challenges, but are models actually learning such knowledge when trained on benchmark tasks? To investigate this, we introduce several new…
We investigate the extent to which modern, neural language models are susceptible to structural priming, the phenomenon whereby the structure of a sentence makes the same structure more probable in a follow-up sentence. We explore how…
Modern reasoning models continue generating long after the answer is already determined. Across five model configurations, two families, and three benchmarks, we find that 52--88% of chain-of-thought tokens are produced after the answer is…
Language models are typically evaluated on their success at predicting the distribution of specific words in specific contexts. Yet linguistic knowledge also encodes relationships between contexts, allowing inferences between word…
Post-training alignment often reduces LLM diversity, leading to a phenomenon known as mode collapse. Unlike prior work that attributes this effect to algorithmic limitations, we identify a fundamental, pervasive data-level driver:…
Previous approaches to turn-taking and response generation in conversational systems have treated it as a two-stage process: First, the end of a turn is detected (based on conversation history), then the system generates an appropriate…
Pre-trained models are nowadays a fundamental component of machine learning research. In continual learning, they are commonly used to initialize the model before training on the stream of non-stationary data. However, pre-training is…
Reasoning models are evaluated on single-turn benchmarks but deployed in multi-turn dialogue, where users push back on correct answers. Under sustained adversarial pressure we find a previously undocumented failure mode: the…
This case study investigates the extent to which a language model (GPT-2) is able to capture native speakers' intuitions about implicit causality in a sentence completion task. We first reproduce earlier results (showing lower surprisal…
We consider the question: when a large language reasoning model makes a choice, did it think first and then decide to, or decide first and then think? In this paper, we present evidence that detectable, early-encoded decisions shape…