feat (tf/pt): add atomic weights to tensor loss

[ChiahsinChu]

Interfaces are of particular interest in many studies. However, the configurations in the training set to represent the interface normally also include large parts of the bulk material. As a result, the final model would prefer the bulk information while the interfacial information is less learnt. It is difficult to simply improve the proportion of interfaces in the configurations since the electronic structures of the interface might only be reasonable with a certain thickness of bulk materials. Therefore, I wonder whether it is possible to define weights for atomic quantities in loss functions. This allows us to add higher weights for the atomic information for the regions of interest and probably makes the model "more focused" on the region of interest.
In this PR, I add the keyword enable_atomic_weight to the loss function of the tensor model. In principle, it could be generalised to any atomic quantity, e.g., atomic forces.
I would like to know the developers' comments/suggestions about this feature. I can add support for other loss functions and finish unit tests once we agree on this feature.

Best.

Summary by CodeRabbit

• New Features

  • Introduced an optional parameter for atomic weights in loss calculations, enhancing flexibility in the TensorLoss class.
  • Added a suite of unit tests for the TensorLoss functionality, ensuring consistency between TensorFlow and PyTorch implementations.

• Bug Fixes

  • Updated logic for local loss calculations to ensure correct application of atomic weights based on user input.

• Documentation

  • Improved clarity of documentation for several function arguments, including the addition of a new argument related to atomic weights.

[coderabbitai]

📝 Walkthrough

📝 Walkthrough

Walkthrough

The changes introduced in this pull request enhance the TensorLoss class in both the deepmd/pt/loss/tensor.py and deepmd/tf/loss/tensor.py files by adding a new boolean parameter enable_atomic_weight. This parameter allows for the optional use of atomic weights during loss calculations. Modifications are made to the __init__ and label_requirement methods to accommodate this feature, and the logic in the forward and build methods is updated to utilize the atomic_weight based on the value of enable_atomic_weight. Additionally, the deepmd/utils/argcheck.py file is updated to improve documentation and include the new argument.

Changes

File Path	Change Summary
deepmd/pt/loss/tensor.py	- Added enable_atomic_weight parameter to __init__. - Updated forward method to use atomic_weight based on enable_atomic_weight. - Modified label_requirement to include DataRequirementItem for atomic_weight.
deepmd/tf/loss/tensor.py	- Added enable_atomic_weight parameter to __init__. - Updated build method to incorporate atomic_weight. - Modified label_requirement to conditionally append DataRequirementItem for atom_weight.
deepmd/utils/argcheck.py	- Updated documentation for doc_global_weight and doc_local_weight. - Added doc_enable_atomic_weight. - Updated loss_tensor function to include enable_atomic_weight.

Possibly related PRs

• feat(pt): support complete form energy loss #3782: The changes in this PR also introduce a new boolean parameter enable_atom_ener_coeff in the EnergyStdLoss and EnergySpinLoss classes, similar to the enable_atomic_weight parameter in the TensorLoss class, indicating a related enhancement in loss calculations involving atomic weights.
• feat(pt): add universal test for loss #4354: This PR introduces a universal test for loss functions, which includes testing for TensorLoss. The modifications in the main PR regarding enable_atomic_weight would likely be relevant for the tests being added here.
• docs: move arg docs from Variant to sub Arguments #4369: While this PR focuses on documentation updates in deepmd/utils/argcheck.py, it mentions the addition of an argument enable_atomic_weight in the loss_tensor function, which is directly related to the changes made in the main PR regarding the same parameter.

Suggested reviewers

• wanghan-iapcm
• njzjz

---

Thank you for using CodeRabbit. We offer it for free to the OSS community and would appreciate your support in helping us grow. If you find it useful, would you consider giving us a shout-out on your favorite social media?

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

🪧 Tips

Chat

There are 3 ways to chat with CodeRabbit:

• Review comments: Directly reply to a review comment made by CodeRabbit. Example:
  • I pushed a fix in commit <commit_id>, please review it.
  • Generate unit testing code for this file.
  • Open a follow-up GitHub issue for this discussion.
• Files and specific lines of code (under the "Files changed" tab): Tag @coderabbitai in a new review comment at the desired location with your query. Examples:
  • @coderabbitai generate unit testing code for this file.
  • @coderabbitai modularize this function.
• PR comments: Tag @coderabbitai in a new PR comment to ask questions about the PR branch. For the best results, please provide a very specific query, as very limited context is provided in this mode. Examples:
  • @coderabbitai gather interesting stats about this repository and render them as a table. Additionally, render a pie chart showing the language distribution in the codebase.
  • @coderabbitai read src/utils.ts and generate unit testing code.
  • @coderabbitai read the files in the src/scheduler package and generate a class diagram using mermaid and a README in the markdown format.
  • @coderabbitai help me debug CodeRabbit configuration file.

Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments.

CodeRabbit Commands (Invoked using PR comments)

• @coderabbitai pause to pause the reviews on a PR.
• @coderabbitai resume to resume the paused reviews.
• @coderabbitai review to trigger an incremental review. This is useful when automatic reviews are disabled for the repository.
• @coderabbitai full review to do a full review from scratch and review all the files again.
• @coderabbitai summary to regenerate the summary of the PR.
• @coderabbitai generate docstrings to generate docstrings for this PR. (Experiment)
• @coderabbitai resolve resolve all the CodeRabbit review comments.
• @coderabbitai configuration to show the current CodeRabbit configuration for the repository.
• @coderabbitai help to get help.

Other keywords and placeholders

• Add @coderabbitai ignore anywhere in the PR description to prevent this PR from being reviewed.
• Add @coderabbitai summary to generate the high-level summary at a specific location in the PR description.
• Add @coderabbitai anywhere in the PR title to generate the title automatically.

CodeRabbit Configuration File (.coderabbit.yaml)

• You can programmatically configure CodeRabbit by adding a .coderabbit.yaml file to the root of your repository.
• Please see the configuration documentation for more information.
• If your editor has YAML language server enabled, you can add the path at the top of this file to enable auto-completion and validation: # yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json

Documentation and Community

• Visit our Documentation for detailed information on how to use CodeRabbit.
• Join our Discord Community to get help, request features, and share feedback.
• Follow us on X/Twitter for updates and announcements.

[coderabbitai]

Actionable comments posted: 0

🧹 Outside diff range and nitpick comments (2)

deepmd/tf/loss/tensor.py (1)

76-79: Consider adding shape validation for atomic_weight

While the implementation is correct, it would be beneficial to validate that the shape of atomic_weight matches the expected dimensions.

 diff = polar - atomic_polar_hat
 diff = tf.reshape(diff, [-1, self.tensor_size])
+# Validate atomic_weight shape
+atomic_weight = tf.ensure_shape(atomic_weight, [tf.shape(diff)[0], 1])
 diff = diff * atomic_weight

deepmd/pt/loss/tensor.py (1)

92-97: Consider adding shape validation for atomic_weight

While the implementation is correct, it would be beneficial to validate the shape of atomic_weight.

 if self.enable_atomic_weight:
     atomic_weight = label["atom_weight"].reshape([-1, 1])
+    # Validate shape matches expected dimensions
+    assert atomic_weight.shape[0] == label["atom_" + self.label_name].reshape([-1, self.tensor_size]).shape[0], \
+        f"Atomic weight shape {atomic_weight.shape} does not match expected shape"
 else:
     atomic_weight = 1.0

📜 Review details

Configuration used: CodeRabbit UI
Review profile: CHILL

📥 Commits

Reviewing files that changed from the base of the PR and between ec3b83f and d53d8af.

📒 Files selected for processing (3)

• deepmd/pt/loss/tensor.py (6 hunks)
• deepmd/tf/loss/tensor.py (3 hunks)
• deepmd/utils/argcheck.py (2 hunks)

🔇 Additional comments (9)

deepmd/tf/loss/tensor.py (3)

43-43: LGTM: Added atomic weight flag

The addition of enable_atomic_weight with a default value of False maintains backward compatibility.

---

70-74: LGTM: Proper atomic weight initialization

The atomic weight is properly initialized either from label_dict or defaults to 1.0 when disabled.

---

174-185: LGTM: Added atomic weight requirement

The atomic weight requirement is properly added to data_requirements when enabled.

deepmd/pt/loss/tensor.py (4)

25-25: LGTM: Well-documented parameter addition

The enable_atomic_weight parameter is properly added with clear documentation explaining its purpose.

Also applies to: 44-45

---

56-56: LGTM: Class initialization

The enable_atomic_weight flag is properly initialized as a class member.

---

116-116: LGTM: Atomic weight application

The atomic weight is correctly applied to the difference calculation.

---

185-195: LGTM: Added atomic weight requirement

The atomic weight requirement is properly added to label_requirement when enabled.

deepmd/utils/argcheck.py (2)

2514-2515: LGTM: Improved documentation clarity

The documentation for global_weight and local_weight has been updated to be more precise and informative.

---

2516-2516: LGTM: Added atomic weight parameter

The enable_atomic_weight parameter is properly added with clear documentation and appropriate default value.

Also applies to: 2528-2534

[codecov]

Codecov Report

Attention: Patch coverage is 88.88889% with 2 lines in your changes missing coverage. Please review.

Project coverage is 84.42%. Comparing base (104fc36) to head (a8e204f).
Report is 1 commits behind head on devel.

Files with missing lines	Patch %	Lines
deepmd/pt/loss/tensor.py	85.71%	1 Missing ⚠️
deepmd/tf/loss/tensor.py	87.50%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##            devel    #4466      +/-   ##
==========================================
+ Coverage   84.39%   84.42%   +0.03%     
==========================================
  Files         670      670              
  Lines       62130    62145      +15     
  Branches     3487     3488       +1     
==========================================
+ Hits        52432    52464      +32     
+ Misses       8571     8554      -17     
  Partials     1127     1127              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[wanghan-iapcm]

Please provide the unit test for the feature.

do you consider implementing it to the pt backend?

[coderabbitai]

Actionable comments posted: 0

🧹 Outside diff range and nitpick comments (3)

source/tests/pt/test_loss_tensor.py (2)

57-57: Simplify dictionary membership test

Instead of if key in np_batch.keys():, you can use if key in np_batch: for better readability and performance.

Apply this diff to implement the change:

-if key in np_batch.keys():
+if key in np_batch:

🧰 Tools

🪛 Ruff (0.8.2)

57-57: Use key in dict instead of key in dict.keys()

Remove .keys()

(SIM118)

---

176-178: Remove unnecessary return statement in tearDown method

The tearDown method should not return a value. Remove the return statement to follow standard unittest practices.

Apply this diff to implement the change:

 def tearDown(self) -> None:
     tf.reset_default_graph()
-    return super().tearDown()
+    super().tearDown()

deepmd/tf/loss/tensor.py (1)

Line range hint 49-49: Fix typo in assertion message

There's a typo in the assertion message: "assian" should be "assign".

Apply this diff to fix the typo:

         ),
-        "Can not assian zero weight both to `pref` and `pref_atomic`"
+        "Can not assign zero weight both to `pref` and `pref_atomic`"
     )

📜 Review details

Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between d53d8af and b4899a1.

📒 Files selected for processing (2)

• deepmd/tf/loss/tensor.py (3 hunks)
• source/tests/pt/test_loss_tensor.py (1 hunks)

🧰 Additional context used

🪛 Ruff (0.8.2)

source/tests/pt/test_loss_tensor.py

57-57: Use key in dict instead of key in dict.keys()

Remove .keys()

(SIM118)

[ChiahsinChu]

do you consider implementing it to the pt backend?

In this PR, this feature is implemented for both tf and pt backend.

[njzjz]

I am not sure if we need to wait for #4105.

[njzjz]

I approve the current PR, but I am not sure when #4105 can be ready.

[wanghan-iapcm]

I approve the current PR, but I am not sure when #4105 can be ready.

I think #4105 should not block the merge of this pr.