https://lzliangzh.github.io Zhong Liang, Peking University en Sat, 03 Oct 2026 06:12:35 +0000 https://lzliangzh.github.io/Blog/2025-12-08-rlhf-notes/ https://lzliangzh.github.io/Blog/2025-12-08-rlhf-notes/ Mon, 08 Dec 2025 00:00:00 +0000 Blog 四篇 RLHF 笔记的合集：策略梯度与目标函数、Actor-Critic 架构、用 Bradley–Terry 建模人类偏好、以及 DPO 的推导。 https://lzliangzh.github.io/Blog/2023-12-31-charge-into-2024/ https://lzliangzh.github.io/Blog/2023-12-31-charge-into-2024/ Sun, 31 Dec 2023 00:00:00 +0000 Blog 又是一年年关岁末。