Closed martinrebane closed 2 years ago
You can set any seed you like, it will only vary the results. You're right that it's just fixed for reproducibility purposes.
On Thu, Dec 16, 2021, 2:27 PM Martin Rebane @.***> wrote:
Hi! I guess seed is fixed in train.py to be able to repeat the results. But just to be sure - the seed is sync'ed for the same reason, syncing is not required to train a new model in SPVNAS or torchsparse?
seed = configs.train.seed + dist.rank( ) configs.workers_per_gpu configs.num_epochs random.seed(seed) np.random.seed(seed) torch.manual_seed(seed) torch.cuda.manual_seed(seed)
If I remove fixed seed from config, I can also remove syncing (unless I need to reproduce the results later)? Thanks!
— Reply to this email directly, view it on GitHub https://github.com/mit-han-lab/spvnas/issues/82, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAVRT3MF7POSLEPPKR4CA53URJDTHANCNFSM5KHHIOAQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.
You are receiving this because you are subscribed to this thread.Message ID: @.***>
Thanks for confirming!
Hi! I guess
seed
is fixed intrain.py
to be able to repeat the results. But just to be sure - theseed
is sync'ed for the same reason, syncing is not required to train a new model in SPVNAS or torchsparse?If I remove fixed seed from config, I can also remove syncing (unless I need to reproduce the results later)? Thanks!