Advances in Large Language Models and Reasoning

The field of large language models (LLMs) and reasoning is rapidly advancing, with a focus on improving the expressive power and robustness of these models. Recent developments have led to the introduction of new techniques, such as in-context learning and chain-of-thought reasoning, which enable LLMs to learn and reason more effectively. Additionally, researchers are exploring the use of human-in-the-loop systems to mitigate the deficiencies of LLMs and improve their performance in risk-sensitive domains. Another area of research is the investigation of subjective factors, such as storytelling, emotions, and hedging, and their impact on argument strength. Overall, the field is moving towards the development of more powerful, robust, and trustworthy LLMs that can be used in a variety of applications. Noteworthy papers include: The paper 'Provable Low-Frequency Bias of In-Context Learning of Representations' which provides a rigorous explanation of the mechanisms by which LLMs achieve in-context learning. The paper 'STITCH: Simultaneous Thinking and Talking with Chunked Reasoning for Spoken Language Models' which introduces a novel generation method that enables simultaneous thinking and talking in spoken language models.

Sources

Provable Low-Frequency Bias of In-Context Learning of Representations

Change of Thought: Adaptive Test-Time Computation

The Emperor's New Chain-of-Thought: Probing Reasoning Theater Bias in Large Reasoning Models

Fail Fast, or Ask: Mitigating the Deficiencies of Reasoning LLMs with Human-in-the-Loop Systems Engineering

Subliminal Learning: Language models transmit behavioral traits via hidden signals in data

STITCH: Simultaneous Thinking and Talking with Chunked Reasoning for Spoken Language Models

Small LLMs Do Not Learn a Generalizable Theory of Mind via Reinforcement Learning

Learning without training: The implicit dynamics of in-context learning

Do Large Language Models Have a Planning Theory of Mind? Evidence from MindGames: a Multi-Step Persuasion Task

R-Stitch: Dynamic Trajectory Stitching for Efficient Reasoning

Investigating Subjective Factors of Argument Strength: Storytelling, Emotions, and Hedging

Built with on top of