Research Papers
Planning and Reasoning
Chain-of-thought, ReAct, process rewards and DeepSeek-R1: the line of work that turned a next-token predictor into something that plans, acts, checks itself, and reasons step by step.
7 chapters53 slides30 minUpdated
Your route through the subject
0/53 slides completed · 0/7 chapters · 0 slides read
Start with: Start Here: Planning and Reasoning, Explained Simply
The chapter map
Follow the route or choose a chapter.- 010/5 completeStart Here: Planning and Reasoning, Explained Simply~2 minReadCode
- 020/8 completeChain of Thought, Teaching Models to Think Out Loud~4 minReadCode
- 030/6 completeReAct, Reasoning and Acting Together~3 minReadField checkPlateCode
- 040/7 completeLet's Verify Step by Step, Rewarding Good Reasoning~3 minReadPlateCode
- 050/7 completeDeepSeek-R1, Learning to Reason Through Reinforcement Learning~3 minReadCode
- 060/7 completeRecursive Language Models, Thinking Beyond the Memory Limit~4 minReadCode
- 070/13 completeGlossary: Planning and Reasoning~11 minRead
Looking for the reading order? Follow this subject’s paper path to see what each paper builds on.
Chapters
Chapter 0: Start Here: Planning and Reasoning, Explained Simply
Five papers trace the climb from a model that answers instantly to one that reasons step by step, acts with tools, checks its work, and learns by practice.
2 min5 slides: Start Here: Planning and Reasoning, Explained Simply0/5 complete
Chapter 1: Chain of Thought, Teaching Models to Think Out Loud
Asking a large model to show its steps before answering, rather than respond instantly, produces a large jump in reasoning accuracy without any retraining.
4 min8 slides: Chain of Thought, Teaching Models to Think Out Loud0/8 complete
Chapter 2: ReAct, Reasoning and Acting Together
ReAct interleaves thought, action, and observation in a loop so a model can call tools for real information, curbing hallucination in modern AI agents.
3 min6 slides: ReAct, Reasoning and Acting Together0/6 complete
Chapter 3: Let's Verify Step by Step, Rewarding Good Reasoning
Grading every step of a model's reasoning, not just the final answer, produces a verifier that rejects lucky guesses and finds correct solutions more often.
3 min7 slides: Let's Verify Step by Step, Rewarding Good Reasoning0/7 complete
Chapter 4: DeepSeek-R1, Learning to Reason Through Reinforcement Learning
DeepSeek-R1 rewards a model only for correct, verifiable answers. Its R1-Zero variant taught itself long reasoning and self-checking with no human examples.
3 min7 slides: DeepSeek-R1, Learning to Reason Through Reinforcement Learning0/7 complete
Chapter 5: Recursive Language Models, Thinking Beyond the Memory Limit
Recursive language models keep a huge input outside the model as explorable data, then write code to inspect pieces and call fresh copies on the rest.
4 min7 slides: Recursive Language Models, Thinking Beyond the Memory Limit0/7 complete
Chapter 6: Glossary: Planning and Reasoning
Chain of thought, self-consistency, process reward models, verifiable rewards, distillation, and inference-time compute are defined for studying reasoning.
11 min13 slides: Glossary: Planning and Reasoning0/13 complete