How to install Unity ML Agents Release 20 in 2023

#unity #mlagents #ArtificialIntelligence #MachineLearning #ReinforcementLearning #AI #NeuralNetworks #gamedev

In this video, I will show you how to install ML-Agents Release 20 for Unity 2022.3 or later, step by step, to train your first Machine Learning project using Unity and Reinforcement Learning.

Before you begin, please follow this guide to install:

00:00 - Intro
00:22 - ML Agents
03:20 - Install Git
04:30 - Clone ML Agents
06:20 - Create Anaconda Env
conda create -n mlagents20 python=3.9
conda activate mlagents20
07:10 - Install PyTorch
07:50 - Install Cuda 11.8
nvcc --version
09:45 - Install ML Agents
pip3 install -e ./ml-agents-envs
pip3 install -e ./ml-agents
mlagents-learn --help
11:10 - Install Protobuf 3.20.3
pip install protobuf==3.20.3
12:20 - Open Unity Project
13:50 - Training PushBlock
15:00 - Install onnx
pip install onnx
15:15 - Re-run Training
16:00 - Test Neural Network

[kushaagra]

ML agents is so difficult to setup correctly. I literally spent hours figuring it out and your tutorial really really helped me out. Thanks a ton! I hope you keep making more vids like these!!!

[leoh]

This is a perfect installation video, showing every possible caveat! Thank you!

[aeQuator]

Dude this is crazy how much u need but not mentioned on the install guide on their page..
Without your video i would've gave up...
Thanks for takeing the time and spreading your knowling..

[charlesAcmen]

omg, i watched your last video for 2022, had trouble and end up with errors, now after this i addressed it!THank you my bro

[SahanRajapaksha-prms]

I have wasted one whole f***ing day until i found your tutorial.Thanks a lot brother 🥰

[well.ginseng]

Thank you for the great video. This is the best video for installing ml-agents.

[mariogravina]

This is excellent. I think one of the biggest issues I'm finding with these installations is the dynamic and not quite compatible/friendly versions of all the components. for instance, v.20 requires Python 3.10 and Torch 1.3.1, however we're at v21 of ML, with Python 3.12 and Torch 1.3.1 (link provided from the ML Installation guide as I follow along) can't find that version, providing instead Torch 2.1. How do you approach these challenges when installing/reinstalling? Like everyone else here (only seven and a half thousand of us) I really appreciate your taking the time to make this video. All the best!

[akella]

Great video! Thanks a lot for this actual tutorial🙏 After trying for launch ml agents due many hours without the result, it was the best solution to watch your video🔥Btw i tried to launch it on py3.11 - didn`t work, then i created another env in Anaconda with py3.9 and repeated all steps again

[charlesAcmen]

i have a question, how can i choose whether cpu training or gpu training?if i have cuda installed?

[johantchassem]

I still have an error after installing the correct version of protobuf . It says: File "<frozen runpy>", line 198, in _run_module_as_main
File "<frozen runpy>", line 88, in _run_code

[snakeywakey]

Hello! I have an AMD gpu, is there any way to utilize it for mlagents? cuda is NVIDIA only.

[HaiPCs]

can anyone help, it keeps giving me the following error:

TypeError: CCompiler_spawn() got an unexpected keyword argument 'env'
[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for numpy
Failed to build numpy
ERROR: Could not build wheels for numpy, which is required to install pyproject.toml-based projects

[CodeHacker-vx]

hii... im getting this error
note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for numpy
Failed to build numpy
ERROR: Could not build wheels for numpy, which is required to install pyproject.toml-based projects @10:31

[GABRIELKENNETHWESLEYWALLACE]

can you help me why when i want to do this mlagents-learn --help its says "module not found error: no module named 'packaging' maybe you can help me, thanks

[eduardorosentreter]

Hola, espero que este bien, quería saber si sabe mas a detalles como ajustar los parámetros de entrenamiento minuto 14:24 se ve en la consola los valores predeterminados supongo, quería saber algunos consejos para ajustar esos parámetros según sea necesario para un modelo en especifico u otro, siempre esta prueba y error, pero creo que se hace engorroso tener que entrenar múltiples veces el mismo modelo en busca de los valores mas ajustados, conoce usted algún truco?

[ryato]

did you know why the preview of the "game" is under 15FPS? i have a RTX3050 i don't understand

[AlMgAgape]

Hi help mee i want to ask, What if i want to implement an algorithm? for example I want to use Deep Q Network for AI self driving car or AI car racing game so where i must to put di DQN algorithm code?

[calalmuradov]

I cannot make it it is writing me that clone is exist and I don't know how to return it

[abdullahdokulmez]

Is NVIDIA Card necessary for this process?

[Ghareebz]

What It Takes To Train a Agent To Shoot ? more Like An enemy that Could Detect The Player And Shoot it.