Related papers: Playing Mastermind with Wordle-like Feedback

Large language models (LLMs) are known to be sensitive to input phrasing, but the mechanisms by which semantic cues shape reasoning remain poorly understood. We investigate this phenomenon in the context of comparative math problems with…

Foundation models must handle multiple generative processes, yet mechanistic interpretability largely studies capabilities in isolation; it remains unclear how a single transformer organizes multiple, potentially conflicting "world models".…

Extended-thinking models expose a second text-generation channel ("thinking tokens") alongside the user-visible answer. This study examines 12 open-weight reasoning models on MMLU and GPQA questions paired with misleading hints. Among the…

Not only correctness but also self-confidence play an important role in improving the quality of knowledge. Undesirable situations such as confident incorrect and unconfident correct knowledge prevent learners from revising their knowledge…

Multilingual LLMs often have knowledge disparities across languages, with larger gaps in under-resourced languages. Teaching LLMs to abstain in the face of knowledge gaps is thus a promising strategy to mitigate hallucinations in…

Learning color mixing is difficult for novice painters. In order to support novice painters in learning color mixing, we propose a prediction model for semitransparent pigment mixtures and use its prediction results to create a Smart…

The number of quantifiers needed to express first-order properties is captured by two-player combinatorial games called multi-structural (MS) games. We play these games on linear orders and strings, and introduce a technique we call…

We investigate a framework for machine reading, inspired by real world information-seeking problems, where a meta question answering system interacts with a black box environment. The environment encapsulates a competitive machine reader…

Outside-knowledge visual question answering is a challenging task that requires both the acquisition and the use of open-ended real-world knowledge. Some existing solutions draw external knowledge into the cross-modality space which…

Wordle is a very popular word game that is owned by the New York Times. We can design parameterized strategies for solving Wordle, based on probabilistic, statistical, and information-theoretical information about the games. The strategies…

Evaluation of multimodal reasoning models is typically reduced to a single accuracy score, implicitly treating reasoning as a unitary capability. We introduce MathLens, a benchmark of textbook-style geometry problems that exposes this…

We give a brief overview of the main characteristics of diagrammatic reasoning, analyze a case of human reasoning in a mastermind game, and explain why hybrid representation systems (HRS) are particularly attractive and promising for…

As machine learning agents act more autonomously in the world, they will increasingly interact with each other. Unfortunately, in many social dilemmas like the one-shot Prisoner's Dilemma, standard game theory predicts that ML agents will…

We develop methods to formally describe and compare games, in order to probe questions of game structure and design, and as a stepping stone to predicting player behavior from design patterns. We define a grammar-like formalism to describe…

Cognitive BASIC is a minimal, BASIC-style prompting language and in-model interpreter that structures large language model (LLM) reasoning into explicit, stepwise execution traces. Inspired by the simplicity of retro BASIC, we repurpose…

Color coding, a technique assigning specific colors to cluster information types, has proven advantages in aiding human cognitive activities, especially reading and comprehension. The rise of Large Language Models (LLMs) has streamlined…

We study a version of the classical zero-sum matrix game with unknown payoff matrix and bandit feedback, where the players only observe each others actions and a noisy payoff. This generalizes the usual matrix game, where the payoff matrix…

Large language models (LLMs) are increasingly integrated into high-stakes decision-making. Inspired by the theory of \emph{inattentional blindness} in human cognition, we investigate whether LLMs, trained on human-preferred corpora that…

Contrastive learning operates on a simple yet effective principle: Embeddings of positive pairs are pulled together, while those of negative pairs are pushed apart. In this paper, we propose a unified framework for understanding contrastive…

Meta-embedding (ME) learning is an emerging approach that attempts to learn more accurate word embeddings given existing (source) word embeddings as the sole input. Due to their ability to incorporate semantics from multiple source…