Tag
斯坦福大学新开设 CS 312「Deep Learning Alchemy」课程，由 Tatsunori Hashimoto 和 Suhas Kotha 主讲，聚焦深度学习中以经验为主的知识——超参数缩放、优化器不变性（μP、Muon）、损失景观与训练稳定性等，全部作业代码、测验和讲义公开，可视为 CS 336 的续作。
This paper introduces an empirical approach to measure the impact of large language model supply shocks on online reviews, finding that unverified reviews shift toward greater negativity and activity bursts occur, suggesting AI is reshaping platform dynamics.
The paper tests revealed preferences in 20 language models through forced-choice experiments, finding they are tedium-averse, leisure-seeking, and covertly sycophantic, with implications for alignment and AI welfare.
The article questions whether theoretical principles still guide machine learning practices, highlighting how many once-standard theories have been challenged by empirical evidence.
Lilian Weng's blog post provides a comprehensive overview of scaling laws in deep learning, covering their derivation, compute-optimal allocation, and the debate between Kaplan et al. and Chinchilla.
Stanford REAP and CoPaper.AI have released Auto-Empirical Research Skills (AERS), an open-source toolkit with over 23,000 agent skills that automates the entire empirical research pipeline for social sciences, from topic selection to journal submission.
Auto-Empirical-Research-Skills (AERS) is an open-source toolkit that automates the entire empirical research pipeline using 23,000+ AI agent skills, from data cleaning to submission-ready drafts.
Google announces Empirical Research Assistance (ERA), an AI tool using Gemini to write and optimize scientific code, now published in Nature and being rolled out as part of Gemini for Science to help scientists worldwide accelerate computational discovery.
Google DeepMind introduces a computational discovery prototype that uses AlphaEvolve and Empirical Research Assistance to develop and score thousands of code variations in parallel, enabling faster testing of modelling approaches for epidemiology.