-
Notifications
You must be signed in to change notification settings - Fork 248
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Tagging Apex version #55
Comments
@benob What os do you use? |
I am using Red Hat Enterprise Linux release 8.1 |
@benob I was using winodws10 and Apex was hard to install. But I can run apex on ubuntu with pytorch 1.7 |
@brightbsit Apex does run with pytorch 1.7, but Oscar doesn't. What commit of Apex did you successfuly complie with pytorch 1.2? |
@benob Now that my computer is broken, i couldn't check my env but I think i used pytorch 1.7 or 1.6. |
@brightbsit @benob I was able to run Oscar with pytorch version 1.7.0+cu101 and latest apex master. |
Do you mean apex@a651e2c24ecf97cbf367fd3f330df36760e1c597? |
@benob I have used the master branch of apex - https://github.com/NVIDIA/apex.git. However, I was able to run oscar code only on a single GPU machine. I had some issues running it on multiple gpus. |
I am able to install apex with torch 1.2.0 and cuda 10.0. I had to use old commit of apex. Here are my commands- git clone https://github.com/NVIDIA/apex.git This commit is from Aug 26 2019. I tried my best in brute force search of compatible commit. If somebody gets a later commit working with OSCAR configuration, please let me know. |
INSTALL.md recommends installing Apex#master. However, Apex does not build any more against pytorch 1.2.0 (see NVIDIA/apex#956). So which version of apex is known to work?
The text was updated successfully, but these errors were encountered: