Build a Doom AI Model with Python | Gaming Reinforcement Learning Full Course

[@adarshd249]

Your TFOD course helped me a lot on doing my project in college. Thank you so much Nick Can't wait to try game reinforcement learning 🔥

[@NicholasRenotte]

Awesome stuff!! Let me know how you go!

[@aymanaslam7267]

This video is great! Really appreciate the time and effort that's been put into making these. Would love to see a series where we could make complex custom environments similar to these pre-made game environments if possible

[@NicholasRenotte]

Definitely working on it @Ayman, I'm not a game dev so it's taking a little time to get up to speed.

[@aymanaslam7267]

@@NicholasRenotte Alright. Thank you so much for all your work on these videos. They really help put a lot of things into perspective from a practical standpoint!

[@markusbuchholz3518]

Impressive project. I do believe that your way in popularization of SW domain, Math and generally knowledge is brilliant. Practical project are extremely helpful to understand quite complex reinforcement learning (or deep reinforcement learning). It it great pleasure to subscribe your channel! Good luck!

[@NicholasRenotte]

Thank you so much @Markus!! I'm definitely enjoying the journey and it's awesome to have you along!

[@luisgongoragaldos7463]

You are a god my man. These kind of videos save my life and reinforce my skills in Python. I am a Doom addict, and I need to improve my Python skills to get a job in Peru. So that helps me as a motivation motor to have the best of two worlds.

[@guilhermetogniolo4221]

Hi Nicholas! I really appreciate the way you explain all the complicated stuff into simple terms... I got all the basics on reinforcement learning and built my first custom env 90% based on your videos \o/. A BIG THANK YOU!

[@netrox25]

Nick deserves more subs tbh 🔥his content is amazing and super helpful!

[@NicholasRenotte]

KZbin algorithm, listen to this guy!!!

[@sharankalyan7]

Every time there's a video from this channel, it has been a game changer for me! Really appreciate the effort :)

[@dimi7083]

I don't normally drop a comment but a big thank you for this video bro a lot of time and effort has gone in and it's much appreciated!

[@NicholasRenotte]

Thanks so much for checking it out dude, truly appreciate it.

[@KenJee_ds]

This is so sick!

[@NicholasRenotte]

Thanks a mil dude!!

[@gq23401]

Thanks for this tutorial. I have been waiting for it since last Friday. Thank bro.

[@NicholasRenotte]

Thanks for checking it out my guy, took a while longer than expected but…it’s here!

[@arturtomasz575]

You inspired me to get back to learn new tools and models outside of my day-to-day stack! Thanks! Btw Marek Wydmuch is pronounced as Marek VydmooH

[@NicholasRenotte]

Thanks a mil @Artur, I knew as soon as I said it I screwed up 😅

[@jindy94]

Late to the party this is such as an awesome video on how RL can be realistically applied to games!! WOW

[@sylvainmazet52]

Awesome video, as usual. Lots of stuff going on, clearly explained. Love it.

[@sylvainmazet52]

I really prefer these to 5 minutes superficial tutorials

[@NicholasRenotte]

Glad you're enjoying it @Sylvain, I definitely prefer doing these over the shorter high level ones as well!

[@flashkachannel2756]

Wow, cool, it is fantastic. I watch for 30 minutes , but I already know that it's a great video

[@NicholasRenotte]

Thanks a mil for checking it out!!

[@cll2598]

Hi Nicholas, thank you very much for the video. When you ran hyperparameter tuning, what was the total timesteps you tuned for, and the number of trials?

[@shyamss2338]

Hi Nicholas, I wanted to know when will a follow up video be released for the Exploratory Data Analysis with the Binance API using Python and Pandas. I have seen that lots of people have also been asking the same question in the Binance API video and I really liked what you did in that video. Hoping you can let us know when the next video in that series is scheduled to come out. Thanks again for everything you do.

[@benmartinez7171]

hi im having some issue with the gym and stable-baselines3 modules, could you share which versions of each you are using?

[@oscarsheen3045]

Hey Nic, how did you come to know that data science was your gig? Did you have times when you thought that programming and data science was a complete drag? Thanks.

[@NicholasRenotte]

Great question Oscar, it’s quite possibly one of the only things I can do for a really long time and not get bored. I’ve done a bunch of different types of coding but there’s something that I find really satisfying about being able to predict stuff. The other cool thing about ds/ml is there is usually more client interaction than just straight software engineering which I really like. Occasionally I do get over it but more often than not it’s because I’m burned out not necessarily bored. When it comes to finding if it’s your thing, I’d highly recommend just jumping into it and giving it a crack!

[@NicholasRenotte]

Holy shit how many times did I say ‘actively’ in the demo of the defend corridor model 🤦‍♂️

[@1p2p1p]

This is a great video. Learnt a ton. Thanks Nicholas, greatly appreciated! I am very interested in seeing a humanoid motion learning video. Seen papers on mujoco based 'Model based RL' . Please look into this topic and create a video.

[@vikashchand.]

Yo! Love this man! 🔥🔥🙌 You reckon we can apply the same concept with games like GTA V and all? 🤔

[@NicholasRenotte]

Heyyyy! So far got street fighter, Minecraft, Trackmania and Rocket League on the hit list!

[@JohnWalz97]

Awesome video man!!! Personal anecdote and not a critique at all, but you should try working on Linux or Mac… Switched from Windows years ago and for dev stuff I could never go back!

[@NicholasRenotte]

Thanks a mil John! I actually went the other way, used to code on MacOS exclusively then realised it was a pain to use Nvidia GPUs on Mac. Still use Linux and RHEL for work but Windows for deep learning on a desktop.

[@JohnWalz97]

@@NicholasRenotte Ah lmao... Yeah that makes sense. Don't do a lot of GPU personally (except for GPU on kubernetes for work) so that's why it works for me 😅

[@youwang9156]

Thank you for ur videos, really appreciate it. Do you think it is possible to make a deep RL model at home to play games like League of Legends and hearthstone ?

[@unknown3.3.34]

Bro What is the prerequisites before looking at this video and where can I learn them? Thank you

[@gustavojuantorena]

Awesome, nick! 👏👏

[@NicholasRenotte]

GUSTAVOOOO thanks so much man!

[@darkwave_2000]

Big Thanks !! First, I always fork the github repo, try to run the script directly, then I watch the details in the video. Safes me some time. Would it be possible to apply the famous open-ai RL algorithm they used for Alpha-GO here as well?

[@NicholasRenotte]

Ooooh, I haven't read into the paper yet, would probably take agesss to train.

[@TeamDman]

Very high quality video! Great job!

[@Powercube7]

Holy hell my man is on a roll

[@NicholasRenotte]

AYYYYYY, my guy! Thank bruv, and thanks for holding down the fort!

[@midimusicforever]

This sounds like a really interesting one!

[@NicholasRenotte]

It was nuts to work on @Midi Music Forever!

[@shamaldesilva9533]

The singularity will come from this channel 😏🤝

[@NicholasRenotte]

😂

[@taocreates]

Any tutorials on learning stability would be great if you would share your finding!

[@haguda4096]

Two points: 1. the link to PPO Algorithm does not work. 2. please more videos on Gaming Reinforcement Learning and best how to program an AI for games like Timberman that don't have helper functions like make_action. Finally, great videos even I from Germany can understand everything because you show very detailed.

[@robkjohnson]

Yooooo…. Ctrl+Enter let’s you run a cell and keep it selected so you don’t have to keep running a cell and pressing up…. Also, very helpful vid lol

[@NicholasRenotte]

This tip just made my day, shift enter be gone. Thanks a mil!!

[@Corianas_]

Is there a way to start with a deathmatch scenario? and having multiple bots in game with just a few in game bots to... spur survival? after they learn to run/hide/heal/hunt, then move to the scenarios going for a goal other than win? But very very interesting thank you for these.

[@NicholasRenotte]

Yeah I think you cool apply curriculum learning to apply increasingly harder levels e.g. more opponents

[@Corianas_]

​@@NicholasRenotte also, while I know why you dropped the doom colours to greyscale, I think you are underestimating how well the game is designed to get the attention to things from a distance. If you instead dropped it to 256 colours I think you would get superior results even if it takes longer to train. (My first laptop had a greyscale screen, and the difference it made in doom going to a colour external monitor was massive, even just small things would pop out when basic VGA(?) colour was applied)

[@RaoSachinYadav1]

It's mind blowing your video you really ML engineer

Great content as always. Could you please make a new video showing how to optimize hyperparameters? I see some articles mention Optuna.

[@NicholasRenotte]

OH man, I wrote the optimization code for this and then thought it wasn't that useful then I had to use it as my model was blowing out. Will show it soon!

@@NicholasRenotte In addition, I would love to learn how to read the Tensorboard charts, to know when the training is going the right direction, or should be stopped and adjust some parametters.

[@aseemk.sharma9531]

Hey @Nicholas Renotte , can we train the agent for deathmatch scenario on similar grounds? I was trying to build an agent but the results were not that good.

[@meetvardoriya2550]

just amazing as always🤩🤩🤩

[@NicholasRenotte]

Heyyyyy! Thanks so much @Meet Vadoria, seriously thanks for always checking out the vids!

[@meetvardoriya2550]

@@NicholasRenotte 😄😄

[@premprasad3511]

What are the prerequisites to learn this ? How much python and ML should I know before starting this ? Any book recommendations on this ?

[@xxfloppypillowxx]

you don't really need ANY to DO this tutorial the issue is that you won't understand much of it. I'd recommend a basic understanding of python as well as a basic understanding of neural networks if you want to actually understand some of it

[@leeshaajane22]

Such a great course!!

[@mohammadhusaynhashemi2274]

Man what a content 😍

[@motivationallines5295]

Excellent work

[@sagara5982]

RL = Renotte learning ur videos r really nice mate!!

[@NicholasRenotte]

Thanks Sagar! Appreciate you checking them out man!

[@lennartlut]

Great video, I learned so much!

[@motivationallines5295]

Hi, Nicholas can we train an AI to play games like pubg or any other FPS game.

[@NicholasRenotte]

From what I've learned it really depends on whether they're integrations into the game servers. As long as we can get some semblence of reward metrics, it's probably possible with enough compute!

[@EmiliaKaida]

Nice tutorial :)

[@vialomur__vialomur5682]

Wow thank you so much!

[@H2nry46]

Hey! It's me again. I just wanted to say that this course is amazing, but you should really slow down while coding so others (such as beginners like me) can * further * understand what you are doing!

[@vasylcf]

Thanks !!!

[@VishnuVardhan-vy1ve]

Bro can I know which algrothim you used

[@prathmeshraut7232]

Sir showing Buffer Szie mismatch

[@sean-mulready]

First, thanks for the video. With this Tutorial I was able to do my first steps in RL as a psychology undergraduate. Only one problem: I couldn't get ViZDoom to be installed on Windows so I set up a dual-boot-system and added Linux. Long story short: As I can't find a suiting alternative to OneNote I'd rather stay with Windows than switching to Linux. Unfortunately I can't find a tutorial on the whole prework (i.e. dependencies) where I seem to mess up. Can you explain or can you recommend a step-by-step-tutorial that I couldn't find so far?

[@joshuastewart4634]

Hello~~ You video help me a lot , thank you~~😄 Could you make full course about how to custom deep reinforcement learning architecture use Open AI Gym and Stable Baseline 3 custom environment ?! Thank you~~

[@mawkuri5496]

can you build a thetan arena ai model tutorial? please? just want to know how to start building with those kind of games.

[@grubersunny3117]

Hi every one did your Agent open fire all the bullets before killing any enemy and causing the mean_reward very low in Defend Center Tutorial

[@maximklechshev6675]

How is it in Sydney?

[@santiaparicio9239]

could it be possible to build an ai for a browser game with selenium? even if there is no official framework? i would love a tutorial of it 💪

[@boop4158]

also how do you set up a working folder? mien all does to appdata\local\programs\python\python310\lib\site-packages ?? I'm not sure what i'm doing... looks nothing like what you have, and you never showed how to set up a working folder nor shown what folders and files you magically had all a sudden.... i'm so confused

[@makhmud_jumanazarov]

thanks this tutorial

[@youssefshoman520]

can you make a tutorial about machine learning on a game that does not have an environment please sir?

[@viswanathansankar3789]

Can you do a video on Contra? My agent keeps spinning.

[@saiganesh2123]

Hi Nicholas how r u Good video 👍🏻

[@NicholasRenotte]

Heyyyy @Sai Ganesh, good good! You?!

[@neerajashish7042]

i guess next project should be teaching AI to play valorant

[@NicholasRenotte]

Oh god, talk about a challenge! Street fighter, trackmania and rocket league are up there!

[@Rhaqim]

Yusss!!!

[@NicholasRenotte]

Ayyy, thanks for checking it out!

[@H2nry46]

Hey there, at 6:38 when I ran the code it showed this "'git' is not recognized as an internal or external command, operable program or batch file." I reviewd the code and it was all correct! Am I missing something?

[@NicholasRenotte]

Try installing git for windows, should be first result in google!

[@H2nry46]

@@NicholasRenotte Ok, thank you! Also, I highly appreciate your high quality videos and your response to almost every comment! 👍

[@boop4158]

you must be missing something.... is this tut outdated? cause it's not working for me, and i'm not sure what you are editing at step 4... i'm soo lost...

[@ffgaming-fe3cx]

subscribed

[@cryptotooth]

This is a great video and very well made, but does anyone know how i can translate the notebook structure to work in something like anaconda? I'm really struggling to understand what code goes where and the order of execution, what's supposed to just be used like a console output etc.

[@ashleysami1640]

Wow wow wow

[@NicholasRenotte]

woah woah woah ;)

[@simonedebellis6783]

[@philtoa334]

:)

[@danielaldana6957]

Remember I wrote to you @twitter inviting you to play something, forget it. You are gonna cheatttttt 🤣😂

[@NicholasRenotte]

😂 gotta play to your strengths man!!

[@danielaldana6957]

@@NicholasRenotte true, I need one of those to replace my old and rusty hand lmao. Come join me @valorant m8

[@NicholasRenotte]

Oh man I can’t game for shit 😂 even my friends bail on me. I end up rpging everything. Maybe once I build a valorant model 🙌🏽

[@PauloHenrique182]

Make it at JAVA, please!

[@Jkauppa]

strafe, mouse move

[@Jkauppa]

mainly strafe

[@Jkauppa]

compute it quickly, record the full movie and then you have the movie to show if wanted

[@Jkauppa]

have noise in your models as parameter so that you dont overfit, and/or do gradient search, not only gradient descent, also ascend in the search, gradient = 0 is the key

[@tiankuangwen]

great videos!very useful to me

[@triggeredcomments4009]

Could make more videos on 3d games please!

[@mrCetus]

Your TFOD course helped me a lot on doing my project in college. Thank you so much Nick Can't wait to try game reinforcement learning 🔥

[@danieladama8105]

Yayyyy!!!!!