Problem of Running distributed version

[Maxwell2017]

@lespeholt Can distributed version use local server instead of GCP?Looking forward to your reply：）

[lespeholt]

Yes. The local script in the readme runs the actors and learner in separate processes.

[Crazyalltnt]

Yes. The local script in the readme runs the actors and learner in separate processes.

Is there a document reference to deploy SEED RL on our own server instead of using GCP, README.md has no relevant information, thank you very much

[Maxwell2017]

The same question as Crazyalltnt. @lespeholt

[lespeholt]

server_address can be changed here: https://github.com/google-research/seed_rl/blob/master/common/common_flags.py#L23 it can also be set to e.g. unix:/tmp/agent_grpc for using Unix domain sockets instead of the network stack.

Other than that, you see the flags applied in the docker file: https://github.com/google-research/seed_rl/blob/master/docker/run.sh#L52

[Pranav-India]

Hi @Crazyalltnt @Maxwell2017 when you refer to local machine do you mean GPU or just a single CPU?

[Maxwell2017]

I mean local servers, not cloud servers

[Pranav-India]

Hi @Maxwell2017 ,ok thank you. do you know if I can run this SEED_RL on a single CPU?

[Maxwell2017]

@Pranav-India Sorry to reply you so late, I think it can run on cpu machine. However, the learner generally runs on gpu. If you use cpu, the distributed acceleration effect may not be achieved.