Please help me out with a subscribe if this video helped you :) Ready for Deep Q-Learning? ruclips.net/video/EUrWGTCGzlA/видео.html My code + list of RL videos in logical order: github.com/johnnycode8/gym_solutions
Thanks for taking the time to make these videos. It's been really hard to find up to date information on how to use this stuff. there's not many videos out there. You break things down very simple and i appreciate it greatly. thanks for the hard work!
Thanks for the great content. I found something interesting about the algorithm. The performance of the algorithm is highly dependent on the results of the epsilon-greedy exploration. If we don't update the informative Q(s,a) table within some episodes (ex.5,000), the results are terrible. It's interesting that the results are inconsistent.
I think VSCode automatically installs the Python extensions when you open a Python file. If that didn’t happen, check out this reference code.visualstudio.com/docs/editor/debugging
Bonjour, je félicite pour ces excellent vidéo, je suis entraîné de programmer le même algorithme mais avec plusieurs agents , c.a.d on a plusieurs agents et plusieurs obstacles et plusieurs gouls au même temps, et j'ai trouve pas une méthode de modifier ce programme et intégrer plusieurs agents au même, merci d'avoir clarifier le programme qui fait cette opération, , et merci bien pour votre aide
The q-table is not accurate at the beginning. It becomes more accurate by updating with the q-learning formula. In the video, I did not talk about the theory and mathematics behind the q-learning formula.
For a project in uni, I want to train an agend that can behave well on different state spaces. Imagine one agend should be able to solve the FrozenLake-Problem in 5x5, but also in 6x6, 7x7 etc. and also 5x6, 5x7, 6x5, etc. How to do that? Do you have an idea or keywords to search for?
My video on how to “Build a Custom Gymnasium Reinforcement Learning Environment” ruclips.net/video/AoGRjPt-vms/видео.html does very similar to what you described. However, you don’t have to create a custom environment, you just have to train the agent on all the different FrozenLake map sizes.
Je vous remercier infiniment , j'ai déjà voir ce vidéo mais j'ai trouvé pas une méthode pour crier plusieurs robots en même , sachat que leur travail est semblable comme le premier agent, tout on évitons les obstacles et fair recherche de la but ( goal), s'il y a une méthode simple merci d'avoir m'informer et le code source surtout , merci et merci pour vous effort de repondre
@@johnnycode merci et merci, puisque m'intéresse au multi agent ( ou bien multi robots ) et plusieurs goal ( buts ) , si l'un des ces agents trouve un goal il le marque comme fait , et lorsque un autre agent trouve le meme goal il le ignore et complet leur travail de recherche , je vous attend, bon implémentation , bon chanse.
The Q-table is a regular Python array, so you can just use a loop to print the value. In my other video, you can visually see the values on the map: ruclips.net/video/1W_LOB-0IEY/видео.html
Please help me out with a subscribe if this video helped you :) Ready for Deep Q-Learning? ruclips.net/video/EUrWGTCGzlA/видео.html
My code + list of RL videos in logical order: github.com/johnnycode8/gym_solutions
Thanks for taking the time to make these videos. It's been really hard to find up to date information on how to use this stuff. there's not many videos out there. You break things down very simple and i appreciate it greatly. thanks for the hard work!
Thanks For putting out the best reinforcement learning tutorial Video I've ever seen. Line by line Brilliant!!!!!!
what a great video. You packed so much in just 12min. Hope you continue to make more videos about RL
Thanks for the great content. I found something interesting about the algorithm. The performance of the algorithm is highly dependent on the results of the epsilon-greedy exploration. If we don't update the informative Q(s,a) table within some episodes (ex.5,000), the results are terrible. It's interesting that the results are inconsistent.
great work
Subscribed and liked. This has been really helpful in getting started. Thank you!
Thanks for this tutorial !
THANKS FOR VIDEO
in visual studio code i don't have button for stop and pause how can I activate them or install any extension
I think VSCode automatically installs the Python extensions when you open a Python file. If that didn’t happen, check out this reference code.visualstudio.com/docs/editor/debugging
@@johnnycode Thank you so much 👍👍👍👍✔✔✔✔
Bonjour, je félicite pour ces excellent vidéo, je suis entraîné de programmer le même algorithme mais avec plusieurs agents , c.a.d on a plusieurs agents et plusieurs obstacles et plusieurs gouls au même temps, et j'ai trouve pas une méthode de modifier ce programme et intégrer plusieurs agents au même, merci d'avoir clarifier le programme qui fait cette opération, , et merci bien pour votre aide
Thank you, good luck on your work.
Based on what do we assign these values to hyperparameters?
Based on trial and error, or a process called hyperparameter tuning.
At the start is the q table accurate?How is the q table made accurate and when does it start to follow it?
The q-table is not accurate at the beginning. It becomes more accurate by updating with the q-learning formula. In the video, I did not talk about the theory and mathematics behind the q-learning formula.
merci d'avoir envoyé le code source de cet vidéo , et merci bien pour ces excellent explication
For a project in uni, I want to train an agend that can behave well on different state spaces. Imagine one agend should be able to solve the FrozenLake-Problem in 5x5, but also in 6x6, 7x7 etc. and also 5x6, 5x7, 6x5, etc.
How to do that? Do you have an idea or keywords to search for?
My video on how to “Build a Custom Gymnasium Reinforcement Learning Environment” ruclips.net/video/AoGRjPt-vms/видео.html does very similar to what you described. However, you don’t have to create a custom environment, you just have to train the agent on all the different FrozenLake map sizes.
Thanks!
Thank you very much!!!
thanks for your explanation
Je vous remercier infiniment , j'ai déjà voir ce vidéo mais j'ai trouvé pas une méthode pour crier plusieurs robots en même , sachat que leur travail est semblable comme le premier agent, tout on évitons les obstacles et fair recherche de la but ( goal), s'il y a une méthode simple merci d'avoir m'informer et le code source surtout , merci et merci pour vous effort de repondre
I will try to do some multiagent videos.
@@johnnycode merci et merci, puisque m'intéresse au multi agent ( ou bien multi robots ) et plusieurs goal ( buts ) , si l'un des ces agents trouve un goal il le marque comme fait , et lorsque un autre agent trouve le meme goal il le ignore et complet leur travail de recherche , je vous attend, bon implémentation , bon chanse.
Bonjour , s'il ya des nouveau pour la programmation des multi agent au meme temps , merci infiniment
what to do to see the Q-table?
The Q-table is a regular Python array, so you can just use a loop to print the value. In my other video, you can visually see the values on the map: ruclips.net/video/1W_LOB-0IEY/видео.html
very clearly!!!
Awesome video
Will you do this with deep Q-learning version ?
Yes, I’m working on it. Will share in a few days.
Thank you !@@johnnycode
Hi, my Deep Q-Learning video is out ruclips.net/video/EUrWGTCGzlA/видео.html
Please check it out.
awesome tutorial more please...
Hi, in case you're looking for a Deep Q-Learning video, I've recently released a detailed one: ruclips.net/video/EUrWGTCGzlA/видео.html