For anyone curious/following this issue it’s been narrowed down to an issue with torch.manual_seed (commenting the manual seeding line out in a3c should make it go away)
1 Like
When I implement A3C with pytorch-gpu, I also encountered this problem. Here is my solution:
add
torch.cuda.manual_seed(args.seed)
mp.set_start_method(‘spawn’) in the start of if name == ‘main’ block
1 Like
Tips to fix-
Restarting the PC worked for some people.
Reduce the resolution. Start with 256 x 256 resolution. Just change the -W 256 -H 256 part in the command.
Try this fork as it requires a lot less VRAM according to many Reddit users.
Regards,
Rachel Gomez