Closed gendrelom closed 2 years ago
Hi @gendrelom Do you have sufficient computing power to run the optimization pipeline? Ray tune is memory and GPU intensive and I ran this on Google Colab (free version). I was able to run through all the trials but yes, I was not able to test more because of computing constraints. Just check if you have good amount of computing or else run in colab
Hi @Athe-kunal I think I have enough computing power for this. I have 32 GB RAM, but Google colab has only 12 GB RAM.
I am talking about VRAM, memory space of GPU. Or else you can set num_gpus = 0 and try again. Maybe the VRAM is not sufficient
But in code num_gpus
is already set to 0, isn't it?
I had the exact same problem some time ago and it turned out to be the machine I was using did not have enough memory.
L
On Jan 25, 2022, at 10:59 AM, Astarag Mohapatra @.***> wrote:
Hi @gendrelom https://github.com/gendrelom Do you have sufficient computing power to run the optimization pipeline? Ray tune is memory and GPU intensive and I ran this on Google Colab (free version). I was able to run through all the trials but yes, I was not able to test more because of computing constraints. Just check if you have good amount of computing or else run in colab
— Reply to this email directly, view it on GitHub https://github.com/AI4Finance-Foundation/FinRL/issues/447#issuecomment-1021460096, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABC5CZNU2JX5RR3PLQOD6TUX3QI3ANCNFSM5MWYL3GQ. You are receiving this because you are subscribed to this thread.
@lcavalie Thanks! How did you fix it? How can I limit the amount of memory to fix this issue on my PC?
@lcavalie Thanks! How did you fix it? How can I limit the amount of memory to fix this issue on my PC?
I was trying to train a model with a 1 minute interval for the entire 2021 year. That was a bit too much I guess. I reduced it to 6 months and everything was fine.
I'm trying to start notebook from this article locally. I slightly modified this notebook and added loop in which I will train a few agents:
But after ~15 minutes after loop start I get this error:
I thought that problem is in Ray, but I can't get this error with Ray examples. What is the problem and how can I fix it?