Blog

I enjoy taking notes on my research and sharing them with others who are also interested in these areas. Thanks to this habit, I have met many like-minded friends who share similar interests.

Large language model

1. Alignment Guidebook.
   
   

   A comprehensive introduction to the preference learning in LLMs.
   
   

2. Why should we do online RLHF/DPO? (in Chinese).
   
   

   A blog about the advantage of online exploration in RLHF.
   
   

3. A tutorial of RLHF with LMFlow (Huggingface version).
   
   

Reinforcement learning theory

1. An introduction to eluder coefficient, a unified framework for nearly all known tractable decision-making problems.
   
   

2. Note on reduction-based RL
   A short note on eluder coefficient, decoupling coefficient, and decision-estimation coefficient.
   
   

3. Note on non-linear contextual bandit
   An application of eluder coefficient in contextual bandit with general function approximation.
   
   

Optimization basis and Decentralized Optimization

1. A generic framework for first-order stochastic decentralized optimization
   
   

2. Analysis of batch gradient descent and stochastic gradient descent (in Chinese)
   
   

Multi-armed bandit basis and Multi-player multi-armed bandit

1. Bandits with stochastic rewards (in Chinese)
   
   

2. Lower bound for stochastic bandit (in Chinese)
   
   

3. Adversarial bandit (in Chinese)
   
   

4. UCB2 (in Chinese)
   
   

5. Introduction to multi-player MAB and Synchronisation Involves Communiation Algorithm
   
   

6. Heterogeneous multi-player MAB
   
   

7. Experiments of MAB
   Also check my research page for the GitHub repository