Papers reading
-
Automated Weak-to-Strong Researcher
June 23, 2026
Anthropic 的「让 AI 自己做对齐研究」实验:用弱模型的标注训练强模型(weak-to-strong),多智能体 swarm 自主提方案、跑实验,弱标签发挥出约 93% 真实标签的能力;记录 AI 给出的策略、reward hacking 案例,以及使用 AI researcher 的经验与未来方向。
Study notes
-
MIT 6.003 Signals and Systems Self-Study Notes
July 9, 2026
Compact self-study notes based on MIT OCW 6.003 Fall 2011: CT/DT LTI systems, convolution, Z/Laplace transforms, frequency response, feedback, Fourier representations, sampling, modulation, and quantization.
-
Matrix Calculus (2)
June 21, 2026
接上篇,整理《Matrix Calculus for Machine Learning and Beyond》第 6–8 章:求根与优化、伴随/反向微分、行列式与逆的导数、正向与反向模式自动微分。
-
Matrix Calculus (1)
June 20, 2026
从看不懂深度学习框架里矩阵乘法的求导讲起,整理《Matrix Calculus for Machine Learning and Beyond》(MIT 18.S096) 的学习笔记。