Skip to content

Conversation

@philokey
Copy link
Contributor

Avoid the problem of training speed become slower iteration by iteration mentioned in this issue

@ICapalija
Copy link
Contributor

In your code learning rate is going to be changed just once, but don't you think that a better approach is to allow multiple corrections of learning rate. Please check my commit c018d83

@philokey
Copy link
Contributor Author

philokey commented Nov 3, 2016

No, it will decay every cfg.TRAIN.STEPSIZE steps with a base of 0.1:
You can get more details here.

@ICapalija
Copy link
Contributor

Yes, you are right. I have been testing the wrong code :(

Copy link
Contributor

@ahmedammar ahmedammar left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have included this commit in my pull request #28

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants