-
Notifications
You must be signed in to change notification settings - Fork 386
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
How long does the training take to achieve a comparable result compared with the paper results? #2
Comments
It takes about 2-3 weeks to train. We train the CenterNet on 8 Tesla V100 (32GB) GPUs and use a batch size of 48. We use the max batch size as we can, because we want to make full use of the memory. The network configuration is similar to CornerNet. If you want to speed up the training time, you may try reducing the batch size and the channels. We are not sure how much worse the result will be with the smaller batch size. |
Or try using the trained model of CornerNet as the pre-trained model |
Can you provide training time in a configuration(including training environment and network configuration)? Cause it takes a long time to achieve a good result for CornerNet.
princeton-vl/CornerNet#67
Many Thanks.
The text was updated successfully, but these errors were encountered: