Tag
This paper introduces VESTA, a framework that equips vision-language models with dynamically growing toolkits for data exploration and statistical model refinement, outperforming prior agent-based methods on complex scientific modeling tasks. The authors also present Dawn, a benchmark for distribution fitting and time series modeling, including real-world astronomy challenges.
A tweet sharing AutoSci, a system from Peking University that automates the entire research lifecycle from literature review to rebuttal, with self-improvement between projects.
Proposes Graphs of Research (GoR), a supervised fine-tuning method that uses citation evolution graphs as supervision for LLM-based research idea generation, achieving state-of-the-art results against gpt-4o-driven baselines.
DeepMind's Deep Tank AI system optimizes the growth of 2D semiconductors, achieving crystals measuring 130 micrometers—surpassing the 100-micrometer target—and significantly accelerating the parameter search process for material fabrication.