4 papers
Wrong Votes, Right Gradients: Why Majority-Vote Rewards Track Verifiable RL
Dylan P. Wilson, Chengsong Huang, Jiaxin Huang
Reinforcement learning with self-generated labels, such as majority vote over a model's own samples, often recovers most of the gain of training with verifiable rewards. This is pu…
Optical Waveguide-based Spider Web Enables Resilient Impact Detection and Localization
Dylan Wilson, Marco Pontin, Peter Walters +1
Spiders use their webs as multifunctional tools that enable capturing and localizing prey and more general environmental sensing through vibrations. Inspired by their biological fu…
LLM Tree Search
Dylan Wilson
This project aims to investigate a novel sequence generation method inspired by the AlphaGo paradigm, adapting it for use with large language models (LLMs). The proposed approach i…
Adjusted Overfitting Regression
Dylan Wilson
In this paper, I will introduce a new form of regression, that can adjust overfitting and underfitting through, "distance-based regression." Overfitting often results in finding fa…