fix(pt): Huber energy loss uses raw model energy

[wanghan-iapcm]

Huber energy loss uses raw model energy instead of energy_pred that includes atom_ener_coeff weighting, so coefficients are silently dropped when use_huber=True

Summary by CodeRabbit

• Bug Fixes

  • Fixed energy Huber loss to use the same precomputed energy values as other loss paths for consistent energy loss computation.
  • Improved energy aggregation handling to be more robust and correctly shaped during computation.

• Tests

  • Expanded loss tests to cover configurations with and without atom energy coefficients and updated test inputs to validate the revised behavior.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

📝 Walkthrough

Walkthrough

Huber branch in energy loss now calls custom_huber_loss with precomputed energy_pred and energy_label tensors instead of model_pred["energy"]/label["energy"]. Tests updated to add enable_atom_ener_coeff flag and related test data fields; minor array-shape/axis clarifications applied in dpmodel loss code.

Changes

Cohort / File(s)	Summary
Energy loss (PyTorch & Paddle) deepmd/pt/loss/ener.py, deepmd/pd/loss/ener.py	Huber-loss branch now passes energy_pred and energy_label to custom_huber_loss instead of using model_pred["energy"] / label["energy"]. Delta, inference conditioning, and aggregation unchanged.
DPModel energy computation deepmd/dpmodel/loss/ener.py	Reshape uses atom_ener.shape and sum uses explicit axis=1 (clarified indexing for robustness).
Tests: consistent energy loss source/tests/consistent/loss/test_ener.py	Test parameterization expanded to (use_huber, enable_atom_ener_coeff); test fixtures now include atom_ener_coeff and find_atom_ener_coeff in predict/label data and propagate enable_atom_ener_coeff into test inputs.

Sequence Diagram(s)

(omitted)

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• feat: add huber loss #4684 — Adjusts Huber-loss energy branch to use precomputed energy_pred/energy_label in the same loss paths.
• feat: dpmodel energy loss & consistent tests #4531 — Modifies energy-loss implementation and related tests touching the same deepmd/*/loss/ener.py and test files.

Suggested reviewers

• njzjz
• iProzd

🚥 Pre-merge checks | ✅ 1 | ❌ 2

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Title check	⚠️ Warning	The title states 'Huber energy loss uses raw model energy', but the actual fix changes it to use precomputed energy_pred instead of raw model energy.	Update title to reflect the actual fix: 'fix(pt): Huber energy loss uses energy_pred with atom_ener_coeff weighting' or similar to accurately describe the correction.
Docstring Coverage	⚠️ Warning	Docstring coverage is 60.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.94%. Comparing base (6c91e99) to head (0fea766).
⚠️ Report is 2 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #5168   +/-   ##
=======================================
  Coverage   81.93%   81.94%           
=======================================
  Files         714      714           
  Lines       73397    73412   +15     
  Branches     3616     3616           
=======================================
+ Hits        60140    60155   +15     
+ Misses      12094    12092    -2     
- Partials     1163     1165    +2     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[iProzd]

Huber energy in pd loss also needs to be fixed.

[iProzd]

xp.shape(atom_ener) should be atom_ener.shape in https://github.com/deepmodeling/deepmd-kit/blob/master/deepmd/dpmodel/loss/ener.py#L134 .

It seems that ut never reached here. @njzjz

[njzjz]

xp.shape(atom_ener) should be atom_ener.shape in

You are correct.