Purpose
This collection keeps factual notes about artificial intelligence: histories of model families, the problems that shaped them, and the design choices that made them practical. The aim is a stable reference for orientation, not opinion pieces or predictions.
Series
- Reinforcement Learning, from First Principles: sixteen notes from what makes RL a distinct learning problem, through the Markov ladder and the Bellman machinery, to a full derivation of policy gradient. The series page carries the reading path; PPO and beyond are the planned next batch.
Planned Areas
- Representation learning and embeddings
- Sequence modeling and attention
- Generative models
- Reinforcement learning foundations (through policy gradient, PPO and beyond still to come)
Status
The reinforcement learning series above is the first complete bilingual set of notes here. New pages should be added in English and Chinese together.