Insights

• Suggestions and thinking on research from J.Schulman
• Randy Pausch Last Lecture: Achieving Your Childhood Dreams
• 用随机梯度下降来优化人生
• 博士这五年

RL_learning

[blogs]| Key Papers in RL

[blogs] | challenges in AI

[Online book] | online RL book

[Online book] | Reinforcement Learning textbook

[Online book] | DeepLearning-500-questions

[videos] | DeepMind X UCL 2021 online course

[videos] | UC berkeley 2020 online course

[videos] | CMU convext Optimization

Paper reading plan

works on speeding up

• Reduce memory usage for LLMs

GPT related

• GPT future
• GPT4 tech report
• AGI

RLHF paper collection repo

• Anthropic LLM
• InstructGPT
• RLHF

Reward shaping

• ICM: Intrinsic Reward -> RND
• Reward shaping
• Policy Dissection
• Pop-Art
• Value Rescale

Deal with POMDP

• PerfectDou
• Perceiver IO
• Gated TransformerXL

code reference

• stable baseline3
• tianshou
• ray
• DL-engine

Topics to follow

• pytorch RL