Add Born Effective Charges support as a property head#1
Open
AugustinLu wants to merge 5 commits intomainfrom
Open
Add Born Effective Charges support as a property head#1AugustinLu wants to merge 5 commits intomainfrom
AugustinLu wants to merge 5 commits intomainfrom
Conversation
|
👋 Jules, reporting for duty! I'm here to lend a hand with this pull request. When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down. I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job! For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with New to Jules? Learn more at jules.google/docs. For security, I will only act on instructions from the user who triggered this task. |
AugustinLu
force-pushed
the
feature/add-bec-support-14624376731387974145
branch
2 times, most recently
from
2e776d7 to
e9656ae
Compare
- Added native support for BEC tensors in the core architecture. - Integrated BEC in calculator for ASE. - Implemented custom loss functions separating diagonal and off-diagonal BEC components. - Added BEC-specific Terminal evaluation metrics (DiagRMSE and OffDiagRMSE). - Introduced new configuration keywords for integration with the 'sevenn' parser. Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
AugustinLu
force-pushed
the
feature/add-bec-support-14624376731387974145
branch
from
e9656ae to
b751faf
Compare
* Explain why the `TotalLoss` tracking metric reads exactly `0.000000` when BEC weights are strictly > 0 and Energy/Force/Stress weights are exactly 0.0 (the terminal logger hardcodes non-BEC features for aggregation visualization).
* Deliver a custom FlashTP compatibility statement affirming that `use_flash_tp: True` flawlessly accelerates the `1x0e+1x1e+1x2e` tensor message-passing layers inside the $L_{max} \ge 2$ environment.
Co-authored-by: AugustinLu <59640670+AugustinLu@users.noreply.github.com>
* Modifies `sevenn/error_recorder.py` to append `BornEffectiveCharges` to the hardcoded loss aggregation list if `KEY.IS_TRAIN_BEC` is enabled in the configuration. * Fixes a bug where models trained exclusively on BECs (`energy/force/stress_weight: 0.0`) reported `TotalLoss = 0.000000`, causing checkpoint saving algorithms tied to `best_metric` improvements to stall. Co-authored-by: AugustinLu <59640670+AugustinLu@users.noreply.github.com>
* **Band Gap**: Adds a global intensive scalar prediction head (`1x0e`) utilizing `AtomReduce` with `reduce='mean'` to prevent artificial scaling across supercells. * **Magmoms**: Adds a node-level scalar prediction head (`1x0e`) mapping to individual atom magnetic moments natively, bypassing spatial derivative calculations. * Creates `BandGapLoss` and `MagmomsLoss` frameworks based off `MSELoss` for direct benchmarking against Materials Project structures. * Outputs properties cleanly through the `SevenNetCalculator`. Co-authored-by: AugustinLu <59640670+AugustinLu@users.noreply.github.com>
Co-authored-by: AugustinLu <59640670+AugustinLu@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Added support for training and predicting Born Effective Charges.
IrrepsLinearto predict a9x0eproperty mapped to(N, 3, 3)per atom matrices.atoms.info/arraysfor BEC extraction.calculator.results.PR created automatically by Jules for task 14624376731387974145 started by @AugustinLu