The Emergence of Abstract Thought in Large Language Models Beyond Any Language
We uncover a core language-agnostic parameter space in large language models—a small but critical set of shared neurons that resemble abstract reasoning beyond any single language.
Jun 11, 2025
ProgressGym: Alignment with a Millennium of Moral Progress
TL;DR. We introduce progress alignment, a new paradigm for aligning AI systems with the trajectory of human moral progress, and provide relevant dataset, benchmark and baseline implementations.
Dec 16, 2024
CAED-Agent: an Agentic Framework to Automate Simulation-Based Experimental Design
TL;DR. We present CAED-Agent, a large-language-model agent framework that utilizes inference-time scaling with lightweight surrogate feedback for cost-aware experimental design.
Sep 27, 2024