Reinforcement Learning 101: 1st out of 6 Sessions

​About Event TL;DR

​AI Scholars 5-week Study Group: Reinforcement Learning 101

​We’re kicking off a 5-week ML study group on Reinforcement Learning 101, one of the next frontiers for building AI systems that can help us make better decisions over time. Reinforcement learning has become an essential piece in modern AI, and it connects beautifully with large language models and emerging world models, making it a great topic if you’re curious about where the field is heading.

​Over five weeks, we’ll be guided by experienced friends who have been working as teaching assistants and researchers in reinforcement learning for the past few years. Each week will be a 1-hour session combining paper reading, conceptual deep dives, and code walkthroughs so you can see how ideas translate into practice. The goal is to make this a very accessible and fruitful learning experience: minimum coding background required, and having some basic machine learning knowledge will be a bonus, not a prerequisite.

​What we’ll cover, 1 hour per wk

• ​Week 1: Introduction to reinforcement learning
  What is reinforcement learning? How is it applied in industry? Why is it important? What are the current trends? We will also walk through the basic math and intuition behind core RL concepts.

• ​Week 2: Value-based RL
  We’ll look at value functions, Q-learning, and related methods, and understand how agents learn to assign value to actions over time.

• ​Week 3: Policy-based RL + combining with value-based RL
  We’ll study policy gradient methods, why they are useful, and how policy-based and value-based approaches can be combined in modern algorithms.

• ​Week 4: RL in Agent: MiniMax's Forge, agent-native RL framework used in training its LLM Model. With comparison to OpenAI reinforcement learning techniques in scalability and agent performance

• ​Week 5: RLHF (Reinforcement Learning from Human Feedback)
  We’ll explore how RLHF is used in aligning large language models and why it has become a critical technique in real-world AI systems.

• ​Week 6: World Model RL
  We’ll discuss world models, model-based RL, and how agents can learn internal models of their environment to plan and generalize better.
  
  

​Who is this for?

• ​Curious learners who want a gentle but serious introduction to reinforcement learning

• ​People with minimal coding experience who are willing to follow walkthroughs

• ​Anyone with a basic understanding of machine learning who wants to see how RL connects with LLMs and world models

​Logistics

• ​Format: Zoom

• ​Schedule: 1 hour per week, for 5 weeks

• ​First session: Saturday, April 4th

• ​Time: 12:00 pm EDT (New York time)

​If you’re excited about the future of decision-making AI and want to explore reinforcement learning in a supportive study group, we’d love to have you join us.

​About AI Scholars

​A peer-led learning journey for engineers, students, researchers, and builders.