Monte Carlo Methods for Model-Free Learning: Part 1
Автор: Priyam Mazumdar
Загружено: 2025-05-30
Просмотров: 360
Code: https://github.com/priyammaz/PyTorch-...
We finally move onto a more practical RL problem: Model-Free Learning! There are a few ways you can solve RL environments without any MDP, but the two main ones are Monte Carlo and TD Learning. Today we explore a simple Monte Carlo implementation!
The idea behind Monte Carlo is very simple. If we don't have an MDP that gives us all the environment information, go ahead and just play the game a ton and then average your returns to estimate the values.
Prereqs are knowing Policy Iteration • Policy Iteration and Value Iteration • Value Iteration !
Timestamps:
00:00:00 - What is Model-Free Learning?
00:04:06 - What is Monte Carlo?
00:07:09 - Monte-Carlo Policy Evaluation
00:09:35 - Trajectories and Returns
00:17:00 - Sample a Trajectory
00:24:50 - Compute Returns
00:30:32 - Estimate Q-Values
00:37:02 - Update the Policy
00:35:50 - Train and Evaluate Model
Socials!
X / data_adventurer
Instagram / nixielights
Linkedin / priyammaz
Discord / discord
🚀 Github: https://github.com/priyammaz
🌐 Website: https://www.priyammazumdar.com/
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
