fix(dpmodel): move econf embeddings to active backend

[njzjz-bot]

Fixes #5641

Summary

• keep electronic-configuration tables as portable NumPy state for serialization
• materialize the table at call time in the namespace, dtype, and device of the first embedding weight
• prevent NativeLayer from selecting NumPy and coercing Torch/JAX/array-API weights back to CPU
• cover tables regenerated by change_type_map() through the same call-time boundary

Why existing tests missed this

The cross-backend consistency tests already include electronic-configuration embeddings, but their JAX, array-api-strict, TF2, and exportable-PyTorch wrappers eagerly convert NumPy attributes when constructing the backend object. That converts both the weights and econf_tebd, so the generic dpmodel mixed-state boundary is never exercised.

The direct dpmodel path uses NumPy parameters and a NumPy electronic-configuration table, where NativeLayer selecting its namespace from the input is correct. Existing tests therefore compared homogeneous NumPy state or homogeneous wrapper-converted state, but never backend-native weights alongside the intentionally portable NumPy constant. CPU-only numerical comparisons also did not assert the returned array backend, dtype, or device.

Validation

• source/tests/consistent/test_type_embedding.py: 64 passed, 49 backend-dependent skips
• focused regression converts only dpmodel layer weights to Torch and verifies Torch output, dtype, device, and NumPy numerical parity
• the same regression repeats after change_type_map() regenerates the NumPy electronic-configuration table
• ruff format .
• ruff check .

Coding agent: Codex
Codex version: codex-cli 0.144.4
Model: gpt-5.6-sol
Reasoning effort: xhigh

Summary by CodeRabbit

• Bug Fixes

  • Electronic-configuration type embeddings now consistently use the active computational backend, data type, and device.
  • Improved compatibility when combining portable model data with PyTorch-based weights.
  • Type-map updates continue to produce consistent results across supported backends.

• Tests

  • Added coverage validating numerical consistency and device/dtype handling for electronic-configuration embeddings.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 3bda8f96-138f-4873-9f37-e94fcc3ea20d

📥 Commits

Reviewing files that changed from the base of the PR and between 6c3b985 and a095055.

📒 Files selected for processing (2)

• deepmd/dpmodel/utils/type_embed.py
• source/tests/consistent/test_type_embedding.py

---

📝 Walkthrough

Walkthrough

TypeEmbedNet.call() now converts electronic-configuration embeddings to the active backend, dtype, and device. A PyTorch-only consistency test validates mixed NumPy constants, PyTorch parameters, and type-map remapping.

Changes

Electronic-configuration backend alignment

Layer / File(s)	Summary
Call-time backend conversion and validation deepmd/dpmodel/utils/type_embed.py, source/tests/consistent/test_type_embedding.py	TypeEmbedNet.call() materializes electronic-configuration embeddings with the active weight’s dtype and device. The PyTorch test verifies output type, device, numerical equivalence, and remapping behavior.

Estimated code review effort: 2 (Simple) | ~10 minutes

Suggested reviewers: njzjz

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly matches the main change: moving electronic-configuration embeddings to the active backend.
Linked Issues check	✅ Passed	The changes implement the requested backend conversion and add regression coverage for backend-native econf embeddings.
Out of Scope Changes check	✅ Passed	The diff stays focused on econf backend materialization and its consistency test, with no obvious unrelated changes.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77.86%. Comparing base (6c3b985) to head (a095055).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #5840      +/-   ##
==========================================
- Coverage   78.58%   77.86%   -0.72%     
==========================================
  Files        1050     1050              
  Lines      120637   120639       +2     
  Branches     4356     4354       -2     
==========================================
- Hits        94801    93936     -865     
- Misses      24278    25144     +866     
- Partials     1558     1559       +1     

☔ View full report in Codecov by Harness.
📢 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.

[njzjz]

Possible reviewers based on changed lines, exact file history, and exact-file review history:

• @OutisLi — 14 reviews on exact changed files (deepmd/dpmodel/utils/type_embed.py).
• @iProzd — 6 commits on changed files (deepmd/dpmodel/utils/type_embed.py, source/tests/consistent/test_type_embedding.py).

No review request was made automatically.

Coding agent: Codex
Codex version: codex-cli 0.144.4
Model: gpt-5.6-sol
Reasoning effort: xhigh