Reinforcement Learning in 3 Hours | Full Course using Python

Want to get started with Reinforcement Learning?
This is the course for you!
This course will take you through all of the fundamentals required to get started with reinforcement learning with Python, OpenAI Gym and Stable Baselines. You'll be able to build deep learning powered agents to solve a varying number of RL problems including CartPole, Breakout and CarRacing as well as learning how to build your very own environment!
In this video you'll learn:
1. All the basics to get up and started with Reinforcement Learning
2. How to build custom environments using OpenAI Gym
3. About working on custom projects for Reinforcement Learning
Get the code for this tutorial: github.com/nicknochnack/Reinf...
Links Mentioned
Stable Baselines 3: stable-baselines3.readthedocs...
OpenAI Gym: gym.openai.com/
PyTorch: pytorch.org/
Atarimania ROMs: www.atarimania.com/roms/Roms.rar
Swig: www.swig.org/Doc1.3/Windows.html
Chapters
0:00 - Start
0:23 - Introduction
1:15 - Gameplan
4:24 - RL in a Nutshell
13:30 - 1. Setup Stable Baselines
21:45 - 2. Environments
30:10 - Loading OpenAI Gym Environments
40:00 - Understanding OpenAI Gym Environments
42:58 - 3. Training
51:32 - Train a Reinforcement Learning Model
1:00:00 - Saving and Reloading Environments
1:04:23 - 4. Testing and Evaluation
1:06:35 - Evaluating RL Models
1:09:34 - Testing the Agent
1:15:56 - Viewing Logs in Tensorboard
1:24:50 - Performance Tuning
1:26:31 - 5. Callbacks, Alternate Algorithms, Neural Networks
1:27:39 - Adding Training Callbacks
1:34:44 - Changing Policies
1:38:27 - Changing Algorithms
1:40:29 - 6. Projects
1:41:31 - Project 1 Atari
1:41:51 - Importing Dependencies
1:44:16 - Applying GPU Acceleration with PyTorch
1:45:11 - Testing Atari Environments
1:51:35 - Vectorizing Environments
1:56:48 - Save and Reload Atari Model
1:57:45 - Evaluate and Test Atari RL Model
2:02:16 - Updated Performance
2:06:34 - Project 2 Autonomous Driving
2:06:56 - Installing Dependencies
2:09:27 - Test CarRacing-v0 Environment
2:12:23 - Train Autonomous Driving Agent
2:17:16 - Save and Reload Self Driving model
2:18:20 - Updated Self Driving Performance
2:28:56 - Project 3 Custom Open AI Gym Environments
2:29:35 - Import Dependencies for Custom Environment
2:32:00 - Types of OpenAI Gym Spaces
2:38:47 - Building a Custom Open AI Environment
2:51:49 - Testing a Custom Environment
2:52:49 - Train a RL Model for a Custom Environment
2:56:22 - Save a Custom Environment Model
2:58:49 - 7. Wrap Up
Oh, and don't forget to connect with me!
LinkedIn: bit.ly/324Epgo
Facebook: bit.ly/3mB1sZD
GitHub: bit.ly/3mDJllD
Patreon: bit.ly/2OCn3UW
Join the Discussion on Discord: bit.ly/3dQiZsV
Happy coding!
Nick
P.s. Let me know how you go and drop a comment if you need a hand!

[@tohbs4062]

Over the summer, I did a research project with a professor at my college over Reinforcement Learning in which both of us were not well versed in the subject. Your videos have helped me out immensely with being able to get a custom environment up.

[@damanpreetsingh8549]

thanks man !! you are a superhero for students learning in the field of A.I !

[@NicholasRenotte]

Thanks sooo much, so glad you're looking into it! @damanpreet!

[@pramanikd]

This is one of the best tutorials on Reinforcement Learning I've come across. Nicholas, your channel is superb.

[@krutyanjayshinde7015]

I cannot tell you lucky I am that I found this channel. Awesome content. Awesome delivery and a overall amazing person. Thanks a lot for putting so much effort in all the videos.

[@ereztison]

Thanks Nicholas for the time and effort you put for creating this great course! it's really doing a great job bridging the gap between theoretical and practical.

[@suryavenkatesan1180]

A perfect Kickstarter for absolute beginners. Thanks for the video!

[@davidjhyatt]

I really appreciate you making this content! Well done and thank you!

[@rixinxie6375]

This is the most charming video for beginer learning RL I have watched so far!! Thank you very much!!

[@siyuancheng9575]

super clear, super beginner friendly and learned tons of super useful tools and info, THANK YOU! Nicholas

[@pavaniddalagi]

Hi Nicholas, thank you very much for your precious time to make us understand the every single step involved in RL. Your tutorials help me a lot. Keep posting such tutorials.🙂

[@jjhj_]

Amazing channel my dude, great quality videos. Your passion for data science makes me really excited to learn a bunch of new things.

[@jveeck33]

What you are doing on your channel is nothing short of awesome. With the expected growth in the field of A.I. and machine learning there's a clear path to this being the most subscribed/watched educational channel on the subject. Thank you and looking forward to what you come out with next

[@NicholasRenotte]

Thanks so much @Joshua, appreciate you checking it out and giving us some love!

[@rhari008]

Excellent explanation on practical approach to RL!! Thanks a lot, Nic! Love from India!

[@victorpessanha2500]

What an amazing content! Took me about a week to go through the whole video 😅Totally worth it though! Appreciate the effort put into this and your other videos as well. Coding is not my strongest point so it's nice to see someone explain everything in detail!

[@yogeshporwal7219]

This is the video we were looking for the long. Thanks a lot!

[@atphamthanh4142]

dude, you are the best. You deserve a medal.

[@NicholasRenotte]

Thanks sooo much man!!

[@riadhbennessib3961]

Thank you so much sir, a compleate tour of the RL tech.

[@pawehajdecki9245]

atari_py is not working anymore instead you need to instal ale_py and then ale-import-roms. Then instead of env.render() you specify render_mode argument in make method env = gym.make('Breakout-v0', render_mode='human')

[@lightstrider001]

If I use render_mode = ‘human’, the model.learn starts the pygame window which wastes a lot of time. Any solution to circumvent that issue. After vectorising, the render function too doesn’t work. Any solutions for this as well?

[@markusbuchholz3518]

Hi Nicholas, I am norwegian so as normal I have some limitations regarding deep language articulation. Therefore I can not sufficiently express my impression about your effort. There is no words in dictionary to formulate your outstanding scarification and willingness to share you passion and knowledge among others. Beside regarding domain is known for me It is always pleasure to hear from you. Yes, from my humble point of view you and your channel are the pearls, who secure only impressive performance. I do need to wish you nothing mores since I know you achieve all your goals! Have a nice day!

[@NicholasRenotte]

Thank you soooo much @Markus!!

[@torarinvik4920]

As a fellow Norwegian, I agree!!! Great content!

[@greymatters1603]

"some limitations regarding deep language articulation"🧐

[@timurnurlygayanov3208]

This is super cool high quality content! the best video about practical start in reinforcement learning I ever saw :) Thank you!

[@ekpopromise2682]

Thanks, so much practical knowledge has been gained. The 3 hours are worth spending

[@monishkumar9190]

Hi Nicholas, this is pure gold. You made it very easy and helpful. Thanks for this. Can you do a video on custom rendering environments please...

[@NicorusG]

Thanks so much Nick, This is awesome well crafted tutorial for introduce me to RL world.

[@etni_dev]

This video is simple amazing. This explains everything from zero knowledge to basics and what to learn next

[@abhishekjn3390]

Man.. I have to start working on ml again.. I watched your sign detection.. Learnt a lot from you.

[@Hassibayub]

Greatest and most complete tutorial ever!

[@creathyang9413]

The content is so great and so helpful, even you charge money for it, I wouldn't mind. Thanks

[@satoshinakamoto5710]

Just finished this course. Thanks Nick! Hopefully we get that render() and pygame tutorial.

[@cyborgx1156]

This video will help me a lot in learning RL. Thanks a lot. It would be great if you can share the PPT presentation.

[@mehadihasan6634]

You and your tutorial really amazing. Please, rapidly share the robotics-based visual custom environment tutorial.

[@felipearenasuribe2546]

Thanks a lot Nic! Excellent guide for beginners. Greetings from Colombia!

[@vikashchand.]

Love this long tutorials man! So much information! 🔥🔥🔥 Drifting skills btw 🤘😂

[@NicholasRenotte]

Ayyyeee, might need to update the model specifically for drifting 😂, i mean I did it to begin with without even trying.

[@lennartlut]

Thank you very much for sharing your knowledge. I really appreciate your effort and I also like your teaching style!

[@guilhermetogniolo4221]

I don't have words to thank you for this AMAZING course. After this video I could finally start walking by myself on machine learning. You are a hero, man... thanks a LOT!!! And you have so many great other videos on your channel... God bless you!!

[@NicholasRenotte]

Thank you soooo much, so glad that you enjoyed it 🙏🙏!!

[@gerasimosfilippakos6585]

Just finished this, I have my major project (RL implemented in a private WoW server to use for Threat Generation) and this helped me A LOT RL all around, and how I should describe my WoW environment to train it. I would love to see live training on a game, but for now I'm planning on simulating the fight entirely into python, and then only using calculated result into the actual game :D

[@NicholasRenotte]

Awesome work @Gerasimos! Sounds like an awesome use case!!

[@saskiadwiulfah2744]

Wow, what a great tutorial! Thank you for your clear explanation Could you please recommend where to start to implement RL in tabular data?

[@xmj8380]

Thank you so much for the video. I'm so lucky to have found your videos when I was getting started in Reinforcement Learning, it helped me so much! It's really great, thank you very, very much!

[@seydaar2065]

Hello Nicholas, your videos are incredibly educational. I would be grateful if you would also post a video for the MuJoCo - Humanoid simulation and make examples of projects.

[@reihanehmirjalili7467]

Thanks so much Nick. your video was amazing and it helped me A LOT! I would be very interested if you make the video you mentioned about training gamed and pygame

[@PedroHenrique-hy7yw]

Greetings from Brazil! I learned a lot with your video, thank you so much!

[@krishnamore2281]

I really love the way you present and explain your codes.

[@NicholasRenotte]

Thanks @Krishna! 🙏🙏🙏

[@sovietcat4825]

this video is so well made and put together, ive gone through so many tutorials and they all use difficult words in the start that confuse me

[@semibenhsan9551]

Just Amazing Nicholas , Thank you very much

[@abhilakshmaheshwari9360]

Awesome video man!!!! Really helpful and to the point :)

[@atharvamadiwale8096]

Heya Nick ! First of all, thank you for the tutorials you have on KZbin. You have saved me from countless assignments tbh😂😂. I also wanted to know if you can do object detection using reinforcement learning. I'm currently exploring the different methods of object detection and wanted to know if it's possible to use reinforcement learning for the same.

[@masoudrahimi3696]

this is great, thank you a lot... please make more videos about RL implementation

[@dragway0385]

It's a shame that this things changes so fast, can be some errors for that, your a great teacher Nich, thank u very much 💖

[@jidushanbojueguitar]

Awesome course!!! This is what I'm expecting!!!

[@ganesh-uc2ft]

Thanks a lot for the amazing comprehensive tutorial. You explained very well and made the implementation so easy. Appreciate the effort you put into it. Thanks again!

[@avdheshkumarsharma5618]

great content and great teaching style...keep making these kind of videos. much love to you brother

[@bextuychiyev7435]

This is a whole HEAP of amazing content :)

[@yunsungchung6083]

SO an amazing tutorial! Thanks a lot

[@kinzhiqing5054]

Thanks for the awesome content learned a lot and very well done simple and easy to understand

[@vendroid6193]

Woah!! Thanks for the content and also for making a full-fledged course for reinforcement learning

[@NicholasRenotte]

Anytime! Let me know what you think of it @VENDROID!

[@vendroid6193]

@@NicholasRenotte It's awesome sir, I learnt so much about reinforcement learning Keep creating such great course, We appreciate your hard work

[@jabed.akhtar]

Thanks thousand time :) ... Best teacher :). The way you made this tutorials is just awesome... Respect for you :)

[@andreasmerz2501]

Daumen hoch, für so ein 3 Stunden Kurs!

[@souravdey1227]

As always I am late to these gold mine of tutorials as well. I believe yours is the only tutorial that should remain in KZbin. That's it. No amount of appreciation feels enough. Having said that, can you please upload some tutorials on Deep Inverse Reinforcement Learning.? Please brother, I beg you.

[@schnookycoder8324]

Thank you so much for this course!!!!❤❤❤❤

[@jkgoodgood1729]

what a wonderful tutorial on RL.!

[@Volpix28]

Great Video! Really appreciate what you´re doing here. An Approach Video of smth like implementing methods against offsetting, preventing side effects, etc. would be cool! Maybe also on how to take an existing environment and implement it within the Gym API to use the Advantages of the Library.

[@user-or7ji5hv8y]

This is really well organized tutorial.

[@NicholasRenotte]

Thanks @C, had a bunchhh of chapters to write out for it!

[@vangelismathioudis3891]

Amazing tutorial Nicholas, thank you for this. I am working on my thesis and I had no prior experience on how to implement code for RL problems.

[@NicholasRenotte]

Awesome work @Vangelis, plenty of new tutorials on how to apply it to gaming on the channel as well! The Doom vid goes int more depth as to how PPO works if you need it.

[@vangelismathioudis3891]

@@NicholasRenotte I am focusing on traffic system control but of course these projects are really interesting and follow them give great hands-on experience.

[@kaihan5962]

very nice video.I hope to see more videos about self-driving in carla

[@ajaykumar-el8vo]

You are superman dude. Amazing teaching. Keep adding more. please help in Consevative QL.

[@henrycroll4827]

Awesome content, thank you so much for putting this together! It might be useful to create a track changes log to account for the updates which have occurred since your original publishing. I was still able to follow most of the tutorial but tracking down those changes took some time.

[@MrClimateCriminal]

guess he is not going to help as he said he would

[@ButchCassidyAndSundanceKid]

Excellent tutorial ! I look forward to spending the whole 3 hours to au fait the knowledge.

[@NicholasRenotte]

Awesome! Still gotta load the weights in for the projects, but will have them up soon!

[@ButchCassidyAndSundanceKid]

@@NicholasRenotte If you want to display the AI Gym on Google Colab, this is the trick. But the refresh rate is terrible, very laggy. !apt-get install python-opengl -y !apt install xvfb -y !pip install pyvirtualdisplay !pip install piglet from pyvirtualdisplay import Display Display().start() import gym from IPython import display import matplotlib.pyplot as plt %matplotlib inline env = gym.make('CartPole-v0') env.reset() img = plt.imshow(env.render('rgb_array')) # only call this once for _ in range(40): img.set_data(env.render('rgb_array')) # just update the data display.display(plt.gcf()) display.clear_output(wait=True) action = env.action_space.sample() env.step(action)

[@kelvinwong9190]

Thank you so much for the Tutorial! watched finish the whole tutorial :) Looking forward to gaming with reinforcement learning :) Cheers!

[@NicholasRenotte]

Yess! So glad you enjoyed it, definitely coming along!

[@uacrazy4481]

So started from scratch, never done this before, didn't even know how to install Python? Anaconda? PyTorch or any of these interesting sounding, I don't even know what to call them. So after a little bit I have managed to get all that going and I am about 40 minutes into your Course and I have managed, I believe at least, as I am still getting some errors about mujoco( haven't installed anything called that yet) but other then that I have managed to figure out my mistakes and am able to make a pole dance back and forth. Amazing. Cannot wait to learn more. Thank you so much for this as even someone who has no idea what even the words mean. Learn this much and have maybe a bit of an understanding. I think it's taken me a good 2 hours to get this far..I am slow but wanted to thank you for the hardwork and time put into this to help even those with a basic understanding. Anyways I hope I can continue to follow along and get to my ultimate goal. Thanks again!! Update.....Literally after I wrote this and continued your Video....You explain how to install Python @ about 50 mins.... I am terrible. Update Okay so since this Video ATARI has updated to Ale so trying to figure out how to do what you did with ATARI but with Ale instead. definitely having issues

[@user-wl8gx6bp9z]

What a tutorial! Thank you so much man!!

[@NicholasRenotte]

Anytime!

[@PacessHO]

Thank you for the great tutorial

[@rethagos8295]

Hey hey Nick, thanks for the material! I'm planning to create a tiny project using reinforcement learning and your videos prove to be a great help for me to get started (although what I seem to need is a model-based algorithm for what I want to do). And as to shutting down Tensorboard from Jupyter - shutting down the process in task manager seems to be good enough. Jupyter will yell at you a bit but no need for kernel shutdown :) saves a lot of hassle

[@NicholasRenotte]

Awesome suggestion @Rethagos!

[@roshanrajprasad]

Thankyou Sir! Really respect you for your work.

[@yasserzeinali4028]

Thanks for this great tutorial.

[@guiliangzheng5704]

Thanks for putting such an awesome class together! Can you share the slides you showed in the class for our reference?

[@gadgetsworld1824]

I got a whole heap of knowledge xD. Thanks alot for the knowledge : )

[@abdelmoula2]

Hello Nicholas, That's for sure the best tutorial i have ever watched, Congrats for the good work. I have a quick question, imagine we want to the setup an observation space like Box but composed of different values ranges. something like V1(from 655 to 700 with int only), V2(from 0 to 14), V3(from -15 to 0). How can we pack this into a Box (or other type of space) that is usable in stable baseline 3. Thank you again.

[@nitroflap]

That's amazing. I was just starting the RL and couldn't find any material on it. Thank you!

[@NicholasRenotte]

Hoping this helps!! Let me know how you go!

[@nitroflap]

@@NicholasRenotte Well, sadly you only covered RL using baselines lib, but would be more helpful if you cover doing so from scratch :D

[@NicholasRenotte]

@@nitroflap I know, it's a huge job though! Will definitely start digging into the algos in the future!

[@nitroflap]

@@NicholasRenotte Looking forward to that. Already learned the DDPG, but would like to understand the A3C, TD3 and others. Anyways, thanks for the tutorial!

[@duchess4738]

Absolutely brilliant thak you

[@mrqwerty7993]

EYOO 100K SUBS LETS GOO !!!! you deserve that

[@noylalzary9092]

Thank you for this amazing video!!!!

[@behradio]

Thanks a lot, It's Amazing ❤️

[@kefasyunana4111]

Thank you for all the great tutorials and intuition. Going through the custom environment, please how do i handle the multiple observations assuming adding the temperature sensor and humidity or pressure sensor data. Thanks

[@mohammadabushanab8703]

I'm here for the 100k, your content is straight firrrrre!

[@NicholasRenotte]

Ayyye, thanks so much @Mohammad!! Much love man!

[@amarsalehi7014]

Hi Nicholas, Thank you for your great video. You mentioned something about finding the optimal structure of ANN using RL. I have searched for it a lot, however, I couldn't find anything useful. Could you please provide us a video regarding this capability or maybe the practical python code to implement that approach?

[@meetvardoriya2550]

Was waiting for this for a long time,thanks a ton ❤️

[@NicholasRenotte]

🙂 anytime @Meet Vardoriya, let me know what you think of it!!

[@meetvardoriya2550]

@@NicholasRenotte gonna sit for 3 hours tommrow and follow up step by step,your videos are very informative!,I really appreciate all your efforts ❤️

[@NicholasRenotte]

@@meetvardoriya2550 awesome! Let me know how you go!!

[@Dave-bz4nr]

First of all, great explanation of the basic concepts. One minor improvement would be to use PEP8 consequently, for instance in string formatting (usage of f-strings instead of str.format).

[@NicholasRenotte]

Solid suggestion @Dave, old habits die hard!

[@user-oi1ri9pt3e]

Awesome, thank you❤

[@vampyrxxx]

You are awesome, thank you so much Nicholas.

[@NicholasRenotte]

Thanks sooo much @Temple of the Ancients! Pumped to have you here!!

[@diegogoujon]

thank you so much !! awesome !

[@manuelvasquez6125]

Thank you for the course. I received a theoretical course but was not really understanding how they were doing the implementation. I feel more comfortable now. I still have 7 days to finish my project, I think Ill make it haha.

[@NicholasRenotte]

You got this! Go getm @Manuel!

[@phoebehuang2689]

Hi Nick, thanks for the turtorial. Would you consider give an example of multi agent learning? Thanks.

[@hirak214]

Thanks a lot, u have been a great help!!

[@ahmedalameldien765]

hi Nicholas , first , thanks for your great videos , you really can't imagine how your tutorials are really helpful to us. second, i want to suggest something, do you how to use RL for applications related to the NLP or CNN ? it would be a great tutorial if you can give us a video of how how to apply reinforcement learning on NLP task (translation, text classification, question answering or summarization) or on CNN task (object detection). i know it is not that easy for us , but i don't think it is impossible for a great mind like yours :)

[@itzmehrshad3433]

thanks for this wonderful course

[@gondorian1640]

Hello Nicholas, great video! Do you perhaps know of any tutorial into the end-to-end implementation? I'm currently doing my msc thesis and it would be helpful if i could learn some ideas on how to implement drl on real environments. Thanks in advance!

[@abdelhassibdad3413]

Hi Nicholas, thank you for this awesome explanation. Can you please drop us the presentation document ?

[@nonen-oo5yk]

Thank you so much for your amazing video

[@MahmoudMohamed-gu2mh]

thanks bro for the nice tutorial

[@baronvonbeandip]

Hey hey, Just chiming in for the pygame custom environment. There's basically nothing out there on the idea and it's exactly what I'm working on. My specific use case, since you asked for specific suggestions, is emulating two agents who have different action sets and rewards that reinforce behavior or decision-making for future interactions in different agent pairs.

[@NoMercy8008]

Ahh NICE, I am looking forward to watching all of those glorious 3 hrs! :D LET'S GO :D Thanks a lot for ALL of that effort, really really appreciate it! :)

[@NicholasRenotte]

YESSSS! Let me know how you go! Super pumped you're checking it out @NoMercy8008!

[@NoMercy8008]

​@@NicholasRenotte Man, I'm such a bonehead :D I just started to write a long off-topic comment suggesting that some of your viewers would benefit of being made aware that tensorflow is also available and easily accessible in javascript without having to install any prerequisites... And then I thought to myself "Maaaaybe, juuust maybe I should check if he has done videos on that already..." and sure enough, there are TONS :D So, yeah... today I learned, once again, that using your brain *before* starting to talk REALLY helps :D And I also learned, once again, that YOU are triple-double-awesome and think of everything! :D So yeah... BUT there's one more thing to add to that: There's also something called ml5.js, which is essentially a layer on top of tensorflow.js, to make it EVEN easier to use and also to provide some more pre-trained and pretty much ready to use models and so on. Pretty nice, and makes it really really easy to get into machine learning :) Anyways, again thanks a lot! Now I'll concentrate on the video :D

[@NoMercy8008]

@@NicholasRenotte (Since this video is quite long, I probably will write multiple comments and split things up a bit, since I had it happen in the past that for example I wrote a SUPER long comment and then scrolled down a bit, which made YT reload the comments to fetch new ones, which deleted my whole novella... live and learn :D So I hope you don't mind, and maybe it even helps you with the YT algorithm, who knows :D) 0:14 HELL YEAH, BROTHER! Mission accomplished, I would say :D (sorry, I just had to!) Also, I noticed that from 1:15 onwards your mic seems to have a very slight bit of audio crackling. (And I don't think it's on my side, at least I don't notice it in other videos or audio files.) It's very minor and really not a big deal, just wanted to make you aware :)

[@NoMercy8008]

@@NicholasRenotte I really find the topic of Neural Network Architecture Search very interesting personally. Choosing the best possible architecture and topology of your network is actually really important and can have huge influence on the outcomes, but it is often not trivial. In many cases you sit there asking yourself "Should I use 2 or 3 dense layers?" or something more complex like "Is it better to use 3 layers with 50 neurons each or two layers with 100?", that kinda stuff. Especially the larger your NNs get, the harder it gets to "just try it out and see", because training can actually take multiple days to fully complete in some cases (probably even more, depending on what you do). So essentially using machine learning to find the best architecture for your machine learning stuff is .. awesome :D Also, I recently played around a lot with NEAT (Neuro-Evolution of Augmenting Topologies; I'm sure I mentioned/suggested this already and many will have already heard about it by now), which is very similiar and probably one of the reasons why I'm so interested with that topic. I mean, you could even see it as "one of many ways to do ML architecture search", since with NEAT you try to evolve some more or less optimal topology over time. Super interesting :) Small suggestion: At around 13:48 you edited a small annotation into the video to correct something. Since the video is 3hrs long, if you miss that anotation, scrolling back to exactly the right moment can be a bit fiddly (and I'm on a desktop pc, I don't want to imagine how that would be with some mobile touch-screen device). So I personally would suggest showing these annotation just a little longer so that most people really can't miss them :)

[@NoMercy8008]

@@NicholasRenotte I love how well-structured all of your videos are, and this one is a perfect example for it! :) With these kinds of non-trivial topics, breaking it down into well-thought-out well-ordered sections really helps building up a great understanding of everything that is important step-by-step, without causing brain overload or anything like that. Devide and conquer! :D Also at the same time, you still show how it actually works, so the actual workflows, how it'll look and feel, what you might run into, etc etc. In my experience, some teachers almost overdo their structuring in such a way that they just have their nice structured slides showing off everything in theory, but they have exactly 0.0f real hands-on examples. And at least for me personally, these actual examples, in other words seeing the teacher actually do it and in a sense getting proof that what the guy talks about is actually real and really works, is very important and helps me enjoy it more :) So, TL;DR: you are a great teacher in my opinion! thanks a lot for that :)