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
Active Learning #610
Active Learning #610
Conversation
All (the pull request submitter and all commit authors) CLAs are signed, but one or more commits were authored or co-authored by someone other than the pull request submitter. We need to confirm that all authors are ok with their commits being contributed to this project. Please have them confirm that by leaving a comment that contains only Note to project maintainer: There may be cases where the author cannot leave a comment, or the comment is not properly detected as consent. In those cases, you can manually confirm consent of the commit author(s), and set the ℹ️ Googlers: Go here for more info. |
@googlebot I consent. 🎉 |
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 creating this! Left some comments. I would also strongly recommend adding script-level testing like you mentioned in the open TODO list so that we can iterate while keeping the script deterministically unchanged (when we want it to).
@dusenberrymw This should now have addressed your comments, contain docstrings and tests. |
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, Joost & Andreas! This LGTM. I'll pull it in and merge.
Three comments/questions:
- Have you both singed the CLA?
- Can you squash the commits to a single commit?
- Also, I believe the commit message for that single commit can be edited such that GitHub will recognize two co-authors if you're interested in having both of you attributed to the commit. I've never done it, but here's the guide.
Uses predictive entropy as acquisition function and a fixed amount of learning steps, for now. Co-authored-by: Andreas Kirsch <blackhc@gmail.com>
|
LGTM as well! Excited to have the code in. |
A UB code style version of the AL notebook. There are several small open todos in the code, but for the majority we (Andreas+me) thought it's time to share and obtain some feedback on the approach.
This script obtains ~80-90% accuracy with ~150 CIFAR-10 samples.
Open todo:
A test followingdeterministic_test.py
.Followpylint
style from the repoAdd docstringsLeft for follow up PR: Proper LR schedule and early stopping - this will improve performance significantly.
Joint work with @BlackHC (Andreas).