-
Notifications
You must be signed in to change notification settings - Fork 68
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
Add io feature to OTF #302
Conversation
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.
Looks good to me, main comment is to make sure the PR description gives a complete summary of the changes (e.g. recording per-species MAEs and version numbers in the output file, recording DFT data and sparse environments, updating the OTF parser, etc.)
Oh yes absolutely, thanks for reviewing and pointing this out! The descriptions are updated |
Co-authored-by: Jonathan Vandermause <[email protected]>
Looks good! Feel free to merge |
Summary
write_model=4
in otf. This mode will back up all checkpoint files after each DFT call. For example, the DFT calls are made at step 1, 11, 23, ..., then there will be foldersckpt_1
,ckpt_11
,ckpt_23
created that in each folder the checkpoint files at that step are stored.In this way, the OTF can restart from any of the previous DFT steps easily with the files and
OTF.from_checkpoint
. This is helpful when the OTF training sometimes gets unphysical after some point in the simulation, e.g. a certain DFT fails to converge but the MD still goes on with the unconverged data, or the gp hyps optimization fails to converge, etc.Checklist
Before a pull request can be merged, the following items must be checked:
Run Black on your local machine.
setup.py
is updated. We are using a version number format a.b.cNote that the CI system will run all the above checks. But it will be much more
efficient if you already fix most errors prior to submitting the PR.