Apple Researchers Propose Cut Cross-Entropy (CCE): A Machine Learning Method that Computes the Cross-Entropy Loss without Materializing the Logits for all Tokens into Global Memory
"语言模型训练" 相关文章
This Machine Learning Research Discusses How Task Diversity Shortens the In-Context Learning (ICL) Plateau
实现机器人领域的ChatGPT时刻，需要大模型+强化学习丨明星教授Sergey特邀报告
How Important is the Reference Model in Direct Preference Optimization DPO? An Empirical Study on Optimal KL-Divergence Constraints and Necessity
