-
Notifications
You must be signed in to change notification settings - Fork 392
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
Notebook for Learning Rate Scheduler #1074
Notebook for Learning Rate Scheduler #1074
Conversation
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks a lot for the PR. I haven't done an in-depth review yet but the first impression is already quite good.
Before proceeding, I have a few points, could you please take a look?
- Right now, the notebook starts with
pip install ...
. We typically have a check if the notebook runs in a colab environment and only run the installs there (also, with a link to open the notebook in colab). E.g. check out this notebook. Could you please copy the approach? - Right now, you apply the
StandardScaler
directly on the data. How about creating an sklearnPipeline
with theStandardScaler
as the first step and the neural net as the 2nd? That would be a bit cleaner IMO. - You can remove the
if __name__ == "__main__":
part. - Regarding the visualization, we have a
simulate
method inLRScheduler
. Maybe that can be used instead of having to implementLRHistoryCallback
. - Could you please add an entry to the notebook/README.md and to the CHANGES.md?
I have modified the notebook, and markdown files as per the suggested changes. Here's a summary:
Could you please review the changes? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks a lot for the updates. I have added some comments, please take a look.
Could you please review the updated changes? |
Could you please review the updated changes? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sorry for the delay. I reviewed the PR again and added a few comments. One change is blocked by a new PR, so I'd say let's wait for that PR first. Please take a look.
add new training method for rate scheduler.
I have addressed the mentioned comments. Could you please review the updated changes? cc: @BenjaminBossan |
Could you please review the updated changes? cc: @BenjaminBossan |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks so much for the updates. I went through the notebook again and found only a few small things I'd like to be adjusted. Please check my comments.
I have addressed the mentioned comments. Could you please review the updated changes? cc: @BenjaminBossan |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for making the final changes, the notebook looks great, well done.
Thank you for assigning me the issue and being a patient guide. Could you please see that #1025 gets closed to declutter? cc: @BenjaminBossan |
Thanks for the reminder! |
As described in #275, I've created a notebook that shows the visualization of the learning rate schedulers using the custom scheduler callback method in skorch.
Please let me know if any improvements are needed and I will make the necessary changes.
Fixes #275
cc: @BenjaminBossan