Question How can I use cloud computing with Unity mlagents ?

I need more processing power and I'm willing to pay for that. How can I run my simulation in cloud computing. Is there any easy way to do this ?

 

Hi,
We don't have anything available at the moment, but it's something we're working on. You can sign up here to be on the mailing list when more details are available.

In the meantime, it's possible to run ML-Agents on something like Google Cloud or AWS. There are some very out-of-date guides for AWS and Azure, but you'll still need to handle things like creating an instance yourself.

 

Thanks for the answer, It might be easier to convert this to gym environment and then use cloud computing maybe ?

 

I'm not sure that would make anything easier, but you're welcome to try it.

 

Yes, you can use cloud computing with Unity mlagents by using a cloud-based gaming platform like Playkey. With Playkey, you can connect to a powerful remote server and run your simulation on that server, using the processing power and resources of the server. This can be a great way to get more processing power and resources for your simulation, and it can also be a great way to save money on your computing costs. If you don't know how to deal with it, you can contact a London IT support company that helped me to run a great server. Good luck!

 

Yes, it is possible to run simulations in cloud computing. There are a few different ways to do this, depending on your specific needs. The most popular way is to use a cloud platform such as Amazon Web Services or Microsoft Azure. These platforms provide a wide range of computing resources and services that you can use to create, configure, and manage your simulation. You can also use a container-based solution such as Docker or Kubernetes to run your simulation on the cloud. These solutions offer the flexibility to scale up or down your computing resources based on your needs. If you're looking for an easier way to get started, you can use a cloud-based simulation platform such as Esyngery cloud computing. This platform provides a web-based interface that you can use to configure, set up, and run your simulations.

 

You can run mlagents in many cloud services by building your unity application as a dedicated server.

To run the dedicated server, and connect it with mlagents-learn, I mean, running those on the same cloud instance is easiest, and I'm going to assume you're doing that, but you still have to connect them together. There are two ways to connect them together:
- approach 1: start mlagents-learn. Run the standalone server with commandline options '--mlagents-port 5004'. Adding this option will make the standalone server try to communciate with the mlagents-learn process.
- approach 2: when you start mlagents-learn, add commandline options --env [path to your dedicated server]. This will auomatically run the server, and get the server to connect with your mlagents-learn process

With approach 2, you can run multiple unity instances, by simply adding also the option '--num-envs [num envs]'. Eg you could do '--num-envs 4', to have 4 parallel unity instances.