Skip to content
This repository has been archived by the owner on Sep 16, 2024. It is now read-only.

the best performance is 0.607 training from scratch while the pretrained model is 0.670 #7

Open
gaopeng-eugene opened this issue Feb 13, 2017 · 2 comments
Open

the best performance is 0.607 training from scratch while the pretrained model is 0.670 #7

gaopeng-eugene opened this issue Feb 13, 2017 · 2 comments

Comments

@gaopeng-eugene
Copy link

No description provided.

@gaopeng-eugene
Copy link
Author

Any idea on how to improve the performance?

@DrSleep
Copy link
Owner

DrSleep commented Feb 21, 2017

Do you use the provided training script for optimisation?
If so, it doesn't follow the choices from the original paper (e.g., Adam vs SGD), which might reflect the difference.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@DrSleep @gaopeng-eugene