feat(pt): allow --init-frz-model for pt model converted from tf model
#5091
Conversation
📝 WalkthroughWalkthroughWhen a frozen PyTorch model is provided, training now attempts non‑strict state_dict loading if an initial strict load raises a RuntimeError about missing or unexpected keys and logs the mismatch. Tests add an end‑to‑end pbtxt→pb→.pth conversion path, UTF‑8 JSON I/O, artifact cleanup, and a new JSON test model. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes
Suggested reviewers
Pre-merge checks and finishing touches❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
📜 Recent review detailsConfiguration used: CodeRabbit UI Review profile: CHILL Plan: Pro 📒 Files selected for processing (3)
🚧 Files skipped from review as they are similar to previous changes (1)
🧰 Additional context used🧠 Learnings (6)📓 Common learnings📚 Learning: 2024-11-25T07:42:55.735ZApplied to files:
📚 Learning: 2025-12-12T13:40:14.334ZApplied to files:
📚 Learning: 2025-12-12T13:40:14.334ZApplied to files:
📚 Learning: 2025-12-12T13:40:14.334ZApplied to files:
📚 Learning: 2025-12-12T13:40:14.334ZApplied to files:
🧬 Code graph analysis (1)source/tests/pt/test_init_frz_model.py (1)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (20)
🔇 Additional comments (5)
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. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (2)
deepmd/pt/train/training.py (2)
619-623: Consider handling "Unexpected key(s)" as well.PyTorch's
load_state_dict(strict=True)can raise RuntimeError for both "Missing key(s)" and "Unexpected key(s)". The current check only handles missing keys, so if the frozen model has extra keys not present inself.model, this would not be caught and would propagate as an unhandled exception.If the intent is to be lenient for any key mismatch, consider a more robust approach:
try: self.model.load_state_dict(frz_model.state_dict()) except RuntimeError as e: - if "Missing key(s) in state_dict" in str(e): + err_msg = str(e) + if "Missing key(s) in state_dict" in err_msg or "Unexpected key(s) in state_dict" in err_msg: self.model.load_state_dict(frz_model.state_dict(), strict=False) - log.warning("Use strict=False to ignore non-matching keys.") log.warning(f"Model state_dict mismatch detected: {e}") + log.warning("Loaded with strict=False to ignore non-matching keys.") else: - raise e + raiseAlternatively, you could use
strict=Falsedirectly and inspect the returnedNamedTuplecontainingmissing_keysandunexpected_keysfor more explicit control.
625-625: Use bareraiseto preserve the original traceback.Per Ruff TRY201, using
raisewithout the exception name preserves the original traceback, which aids debugging.else: - raise e + raise
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
deepmd/pt/train/training.py(1 hunks)
🧰 Additional context used
🧠 Learnings (2)
📓 Common learnings
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
📚 Learning: 2024-11-25T07:42:55.735Z
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
Applied to files:
deepmd/pt/train/training.py
🧬 Code graph analysis (1)
deepmd/pt/train/training.py (1)
deepmd/pd/train/wrapper.py (2)
load_state_dict(186-191)state_dict(199-203)
🪛 Ruff (0.14.8)
deepmd/pt/train/training.py
625-625: Use raise without specifying exception name
Remove exception name
(TRY201)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (31)
- GitHub Check: Build C++ (clang, clang)
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build wheels for cp311-macosx_x86_64
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Build C++ (cpu, cpu)
- GitHub Check: Build C++ (cuda, cuda)
- GitHub Check: Build wheels for cp311-win_amd64
- GitHub Check: Build wheels for cp311-macosx_arm64
- GitHub Check: Build C++ (rocm, rocm)
- GitHub Check: Build C++ (cuda120, cuda)
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Test Python (5, 3.12)
- GitHub Check: Test Python (6, 3.12)
- GitHub Check: Test Python (5, 3.9)
- GitHub Check: Test Python (4, 3.9)
- GitHub Check: Test Python (4, 3.12)
- GitHub Check: Test Python (3, 3.9)
- GitHub Check: Test Python (3, 3.12)
- GitHub Check: Test Python (6, 3.9)
- GitHub Check: Test Python (1, 3.12)
- GitHub Check: Test Python (2, 3.12)
- GitHub Check: Test Python (1, 3.9)
- GitHub Check: Test C++ (false, false, false, true)
- GitHub Check: Test Python (2, 3.9)
- GitHub Check: Build C library (2.14, >=2.5.0,<2.15, libdeepmd_c_cu11.tar.gz)
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
- GitHub Check: Test C++ (true, true, true, false)
- GitHub Check: Test C++ (true, false, false, true)
- GitHub Check: Test C++ (false, true, true, false)
- GitHub Check: Analyze (python)
- GitHub Check: Analyze (c-cpp)
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## devel #5091 +/- ##
=======================================
Coverage 82.14% 82.14%
=======================================
Files 709 709
Lines 72453 72457 +4
Branches 3615 3615
=======================================
+ Hits 59514 59519 +5
Misses 11775 11775
+ Partials 1164 1163 -1 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (2)
source/tests/pt/test_init_frz_model.py (2)
137-163: Consider adding assertions to strengthen the test.While this smoke test verifies that the TF→PT conversion and initialization doesn't crash (which aligns with the PR objective), it could be enhanced with basic assertions to verify the model loaded successfully.
For example, you could add verification that the trainer was created successfully:
trainer = get_trainer(config, init_frz_model=frozen_model) assert trainer is not None, "Trainer should be created successfully" trainer.run()Or verify that the frozen model file exists before using it:
assert Path(frozen_model).exists(), f"Converted model {frozen_model} should exist" trainer = get_trainer(config, init_frz_model=frozen_model)
163-163: Remove trailing whitespace.Line 163 appears to have trailing whitespace.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
deepmd/pt/train/training.py(1 hunks)source/tests/pt/test_init_frz_model.py(5 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- deepmd/pt/train/training.py
🧰 Additional context used
🧠 Learnings (1)
📓 Common learnings
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
🧬 Code graph analysis (1)
source/tests/pt/test_init_frz_model.py (3)
deepmd/entrypoints/convert_backend.py (1)
convert_backend(11-31)deepmd/pt/entrypoints/main.py (1)
get_trainer(98-206)deepmd/pt/train/training.py (1)
run(716-1212)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (31)
- GitHub Check: Test Python (5, 3.12)
- GitHub Check: Test Python (4, 3.12)
- GitHub Check: Test Python (6, 3.9)
- GitHub Check: Test Python (3, 3.9)
- GitHub Check: Test Python (1, 3.12)
- GitHub Check: Test Python (6, 3.12)
- GitHub Check: Test Python (3, 3.12)
- GitHub Check: Test Python (4, 3.9)
- GitHub Check: Test Python (5, 3.9)
- GitHub Check: Test Python (2, 3.9)
- GitHub Check: Test Python (1, 3.9)
- GitHub Check: Test Python (2, 3.12)
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
- GitHub Check: Build C library (2.14, >=2.5.0,<2.15, libdeepmd_c_cu11.tar.gz)
- GitHub Check: Analyze (python)
- GitHub Check: Test C++ (false, true, true, false)
- GitHub Check: Test C++ (true, true, true, false)
- GitHub Check: Test C++ (false, false, false, true)
- GitHub Check: Test C++ (true, false, false, true)
- GitHub Check: Analyze (c-cpp)
- GitHub Check: Build C++ (cuda120, cuda)
- GitHub Check: Build C++ (clang, clang)
- GitHub Check: Build C++ (cpu, cpu)
- GitHub Check: Build C++ (cuda, cuda)
- GitHub Check: Build C++ (rocm, rocm)
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build wheels for cp311-win_amd64
- GitHub Check: Build wheels for cp311-macosx_x86_64
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Build wheels for cp311-macosx_arm64
🔇 Additional comments (4)
source/tests/pt/test_init_frz_model.py (4)
23-28: LGTM!The new imports are appropriate for testing the TensorFlow to PyTorch model conversion pathway.
37-37: LGTM!Explicit UTF-8 encoding improves cross-platform compatibility.
66-66: LGTM!Explicit UTF-8 encoding improves cross-platform compatibility.
174-175: LGTM!Proper cleanup of test artifacts (.pb files) ensures tests don't leave behind intermediate files.
There was a problem hiding this comment.
Actionable comments posted: 0
♻️ Duplicate comments (1)
source/tests/pt/test_init_frz_model.py (1)
151-151: Remove redundant assignment.This assignment is overridden by line 160, making it dead code.
Apply this diff to remove the redundant line:
- config["training"]["numb_steps"] = 1 config["training"]["save_freq"] = 1
🧹 Nitpick comments (1)
source/tests/pt/test_init_frz_model.py (1)
146-146: Consider converting the comment to a docstring.The comment describing the test purpose would be better placed as a method docstring for improved documentation and discoverability.
Apply this diff:
def test_init_frz_model_pb2pth(self) -> None: + """Test initialization from frozen model with mismatched keys to test strict=False fallback.""" frozen_model = "frozen_model_pb2pth.pth" # Convert pth model from pb model convert_pbtxt_to_pb( str(Path(__file__).parent / "water/se_e2_a.pbtxt"), "frozen_model.pb" ) convert_backend(INPUT="frozen_model.pb", OUTPUT=frozen_model) - """Test initialization from frozen model with mismatched keys to test strict=False fallback.""" # Create a base model
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
source/tests/pt/test_init_frz_model.py(5 hunks)
🧰 Additional context used
🧠 Learnings (1)
📓 Common learnings
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
🧬 Code graph analysis (1)
source/tests/pt/test_init_frz_model.py (1)
deepmd/entrypoints/convert_backend.py (1)
convert_backend(11-31)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (30)
- GitHub Check: Build C++ (cpu, cpu)
- GitHub Check: Build C++ (rocm, rocm)
- GitHub Check: Build C++ (clang, clang)
- GitHub Check: Build C++ (cuda120, cuda)
- GitHub Check: Build C++ (cuda, cuda)
- GitHub Check: Test Python (5, 3.9)
- GitHub Check: Test Python (6, 3.9)
- GitHub Check: Test Python (5, 3.12)
- GitHub Check: Test Python (6, 3.12)
- GitHub Check: Test Python (4, 3.12)
- GitHub Check: Test Python (2, 3.12)
- GitHub Check: Test Python (4, 3.9)
- GitHub Check: Test Python (1, 3.9)
- GitHub Check: Test Python (2, 3.9)
- GitHub Check: Test Python (3, 3.12)
- GitHub Check: Test Python (3, 3.9)
- GitHub Check: Test Python (1, 3.12)
- GitHub Check: Test C++ (false, true, true, false)
- GitHub Check: Test C++ (true, false, false, true)
- GitHub Check: Test C++ (true, true, true, false)
- GitHub Check: Test C++ (false, false, false, true)
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build wheels for cp311-macosx_x86_64
- GitHub Check: Build wheels for cp311-win_amd64
- GitHub Check: Build wheels for cp311-macosx_arm64
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
- GitHub Check: Build C library (2.14, >=2.5.0,<2.15, libdeepmd_c_cu11.tar.gz)
- GitHub Check: Analyze (c-cpp)
🔇 Additional comments (4)
source/tests/pt/test_init_frz_model.py (4)
16-28: LGTM! Imports support the new test functionality.The imports for
convert_backendandconvert_pbtxt_to_pbare necessary for the new test case that exercises model conversion from TensorFlow to PyTorch formats.
38-38: LGTM! UTF-8 encoding improves cross-platform compatibility.Explicitly specifying UTF-8 encoding for JSON file operations is a best practice that prevents encoding issues across different platforms and locales.
Also applies to: 67-67
138-163: Verify test coverage: no assertions present.The test method successfully exercises the conversion and initialization pathway but includes no assertions to validate correctness. Given the PR objectives focus on preventing crashes when loading converted models, this may be intentional—the test passes if no exceptions are raised.
Consider adding assertions if specific behavior should be validated beyond "does not crash":
# Example: verify the trainer was initialized successfully self.assertIsNotNone(trainer) # Or validate that the model can perform inference # (similar to test_dp_test method)
175-176: LGTM! Proper cleanup of test artifacts.The tearDown correctly removes the generated
.pbfiles, preventing test pollution and maintaining a clean test environment.
ChiahsinChu
force-pushed
the
devel-init-from-tf2pt
branch
from
b6c525a to
c96476a
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
source/tests/pt/test_init_frz_model.py (1)
138-162: Consider adding assertions to verify test behavior.The test verifies that initialization from a TensorFlow-converted model doesn't crash, but it lacks assertions to confirm the model was properly initialized. While the implicit "doesn't crash" check is valuable, adding explicit verification would make the test more robust and its intent clearer.
Consider adding assertions after line 162, such as:
# Verify the trainer was initialized successfully assert trainer is not None assert trainer.model is not None # Optionally verify model can perform inference freeze(model="model.pt", output="test_frozen.pth", head=None) dp = DeepPot("test_frozen.pth") # Perform a simple inference check similar to test_dp_test
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
deepmd/pt/train/training.py(1 hunks)source/tests/pt/test_init_frz_model.py(5 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- deepmd/pt/train/training.py
🧰 Additional context used
🧠 Learnings (1)
📓 Common learnings
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
🧬 Code graph analysis (1)
source/tests/pt/test_init_frz_model.py (3)
deepmd/entrypoints/convert_backend.py (1)
convert_backend(11-31)deepmd/pt/entrypoints/main.py (2)
main(521-580)freeze(372-386)deepmd/main.py (1)
main(973-1020)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (20)
- GitHub Check: Analyze (javascript-typescript)
- GitHub Check: Analyze (c-cpp)
- GitHub Check: Analyze (python)
- GitHub Check: Test C++ (true, false, false, true)
- GitHub Check: Test C++ (false, true, true, false)
- GitHub Check: Test C++ (false, false, false, true)
- GitHub Check: Test C++ (true, true, true, false)
- GitHub Check: Build C library (2.14, >=2.5.0,<2.15, libdeepmd_c_cu11.tar.gz)
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
- GitHub Check: Build wheels for cp311-macosx_x86_64
- GitHub Check: Build wheels for cp311-win_amd64
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build wheels for cp311-macosx_arm64
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build C++ (rocm, rocm)
- GitHub Check: Build C++ (cuda, cuda)
- GitHub Check: Build C++ (clang, clang)
- GitHub Check: Build C++ (cpu, cpu)
- GitHub Check: Build C++ (cuda120, cuda)
🔇 Additional comments (4)
source/tests/pt/test_init_frz_model.py (4)
16-18: LGTM!The import is correctly added and properly used in the new test method at line 144.
26-28: LGTM!The import is correctly added and used in the new test method to convert pbtxt models to pb format.
38-38: LGTM!Explicitly specifying UTF-8 encoding is good practice for ensuring consistent file I/O behavior across different platforms.
Also applies to: 67-67
174-175: LGTM!The cleanup logic correctly handles the .pb files created by the new test method, ensuring test artifacts don't persist.
ChiahsinChu
force-pushed
the
devel-init-from-tf2pt
branch
from
c96476a to
4069606
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
source/tests/pt/test_init_frz_model.py (2)
66-71: Fix temp file handle leak + delete the temp JSON (can break on Windows).
NamedTemporaryFile(delete=False, ...)stays open and the.jsonis never removed; also reopening an open temp file can fail on Windows. Suggest writing via the temp handle, closing it, andos.removeinfinally.- tmp_input = tempfile.NamedTemporaryFile(delete=False, suffix=".json") - with open(tmp_input.name, "w", encoding="utf-8") as f: - json.dump(empty_config, f, indent=4) - run_dp( - f"dp --pt train {tmp_input.name} --init-frz-model {self.models[-1]} --use-pretrain-script --skip-neighbor-stat" - ) + tmp_input = tempfile.NamedTemporaryFile( + delete=False, suffix=".json", mode="w", encoding="utf-8" + ) + try: + json.dump(empty_config, tmp_input, indent=4) + finally: + tmp_input.close() + try: + run_dp( + f"dp --pt train {tmp_input.name} --init-frz-model {self.models[-1]} --use-pretrain-script --skip-neighbor-stat" + ) + finally: + os.remove(tmp_input.name)
164-175: TearDown cleanup: also removese_e2_a.hdf5(likely produced by new JSON).
Right now.pbcleanup is added (good), but iftraining.stat_fileis honored,./se_e2_a.hdf5may remain and cause test pollution.for f in os.listdir("."): @@ if f.startswith("frozen_model") and f.endswith(".pb"): os.remove(f) + if f in ["se_e2_a.hdf5"]: + os.remove(f)
🧹 Nitpick comments (1)
source/tests/pt/test_init_frz_model.py (1)
138-163: Nice: pb→pth init coverage; consider asserting conversion outputs exist before training.
A quickPath(...).is_file()check beforeget_trainer(...)makes failures more local/easier to diagnose if conversion breaks.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
deepmd/pt/train/training.py(1 hunks)source/tests/pt/model/models/se_e2_a.json(1 hunks)source/tests/pt/test_init_frz_model.py(5 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- deepmd/pt/train/training.py
🧰 Additional context used
🧠 Learnings (1)
📓 Common learnings
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
🧬 Code graph analysis (1)
source/tests/pt/test_init_frz_model.py (3)
deepmd/entrypoints/convert_backend.py (1)
convert_backend(11-31)deepmd/pt/entrypoints/main.py (2)
main(521-580)freeze(372-386)deepmd/main.py (1)
main(973-1020)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (31)
- GitHub Check: Test Python (5, 3.9)
- GitHub Check: Test Python (6, 3.12)
- GitHub Check: Test Python (5, 3.12)
- GitHub Check: Test Python (1, 3.9)
- GitHub Check: Test Python (4, 3.12)
- GitHub Check: Test Python (4, 3.9)
- GitHub Check: Test Python (6, 3.9)
- GitHub Check: Test Python (1, 3.12)
- GitHub Check: Build wheels for cp311-macosx_arm64
- GitHub Check: Test Python (2, 3.9)
- GitHub Check: Test Python (2, 3.12)
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Test Python (3, 3.12)
- GitHub Check: Build wheels for cp311-macosx_x86_64
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build wheels for cp311-win_amd64
- GitHub Check: Test Python (3, 3.9)
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
- GitHub Check: Build C library (2.14, >=2.5.0,<2.15, libdeepmd_c_cu11.tar.gz)
- GitHub Check: Analyze (python)
- GitHub Check: Analyze (c-cpp)
- GitHub Check: Build C++ (cpu, cpu)
- GitHub Check: Build C++ (cuda120, cuda)
- GitHub Check: Build C++ (rocm, rocm)
- GitHub Check: Build C++ (cuda, cuda)
- GitHub Check: Build C++ (clang, clang)
- GitHub Check: Test C++ (false, false, false, true)
- GitHub Check: Test C++ (true, false, false, true)
- GitHub Check: Test C++ (true, true, true, false)
- GitHub Check: Test C++ (false, true, true, false)
🔇 Additional comments (2)
source/tests/pt/test_init_frz_model.py (2)
37-43: UTF-8 I/O: LGTM.
Consistent explicit encoding helps avoid CI locale surprises.Also applies to: 67-68
16-28: API stability concern is unfounded; both functions are established and stable.
convert_pbtxt_to_pbis a documented TensorFlow utility already widely used across TF tests, andconvert_backendis a public CLI subcommand with stable signature used in production model code. Both APIs are well-established and appropriate test dependencies. TensorFlow requirement forconvert_pbtxt_to_pbis expected behavior for protobuf text conversion, not a new concern.
ChiahsinChu
force-pushed
the
devel-init-from-tf2pt
branch
from
4069606 to
1be559c
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
source/tests/pt/test_init_frz_model.py (1)
138-162: Add assertions to validate successful initialization from the converted pb→pth model.The test verifies that initialization doesn't crash but lacks assertions to confirm the model was properly initialized. The
test_dp_testmethod in the same file demonstrates appropriate validation patterns—consider adding similar inference checks, such as creating aDeepPotinstance from the initialized model and performing a basic evaluation to confirm state transfer was successful.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
deepmd/pt/train/training.py(1 hunks)source/tests/pt/model/models/se_e2_a.json(1 hunks)source/tests/pt/test_init_frz_model.py(5 hunks)
✅ Files skipped from review due to trivial changes (1)
- source/tests/pt/model/models/se_e2_a.json
🚧 Files skipped from review as they are similar to previous changes (1)
- deepmd/pt/train/training.py
🧰 Additional context used
🧠 Learnings (1)
📓 Common learnings
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (31)
- GitHub Check: Test Python (4, 3.9)
- GitHub Check: Test Python (6, 3.12)
- GitHub Check: Test Python (4, 3.12)
- GitHub Check: Test Python (5, 3.12)
- GitHub Check: Test Python (6, 3.9)
- GitHub Check: Test Python (5, 3.9)
- GitHub Check: Test Python (1, 3.9)
- GitHub Check: Test Python (1, 3.12)
- GitHub Check: Test Python (2, 3.9)
- GitHub Check: Test Python (3, 3.12)
- GitHub Check: Test Python (2, 3.12)
- GitHub Check: Test Python (3, 3.9)
- GitHub Check: Build C++ (cuda120, cuda)
- GitHub Check: Build C++ (rocm, rocm)
- GitHub Check: Build C++ (cuda, cuda)
- GitHub Check: Build C++ (clang, clang)
- GitHub Check: Build C++ (cpu, cpu)
- GitHub Check: Test C++ (false, true, true, false)
- GitHub Check: Test C++ (true, true, true, false)
- GitHub Check: Analyze (c-cpp)
- GitHub Check: Test C++ (true, false, false, true)
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Analyze (python)
- GitHub Check: Build wheels for cp311-macosx_arm64
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Build wheels for cp311-win_amd64
- GitHub Check: Build wheels for cp311-macosx_x86_64
- GitHub Check: Test C++ (false, false, false, true)
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build C library (2.14, >=2.5.0,<2.15, libdeepmd_c_cu11.tar.gz)
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
🔇 Additional comments (4)
source/tests/pt/test_init_frz_model.py (4)
16-18: LGTM!The import is properly used in the new test method to convert frozen models from pb to pth format.
26-28: LGTM!The import is properly used in the new test method to convert pbtxt models to pb format.
38-38: LGTM!Explicitly specifying UTF-8 encoding is a best practice for JSON files and ensures consistent behavior across different platforms.
Also applies to: 67-67
174-175: LGTM!The cleanup correctly removes the
.pbfiles generated during the pb→pth conversion test, maintaining proper test isolation.
ChiahsinChu
force-pushed
the
devel-init-from-tf2pt
branch
from
1be559c to
5d2e44d
Compare
There was a problem hiding this comment.
Actionable comments posted: 1
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
deepmd/pt/train/training.py(1 hunks)source/tests/pt/model/models/se_e2_a.json(1 hunks)source/tests/pt/test_init_frz_model.py(5 hunks)
🚧 Files skipped from review as they are similar to previous changes (2)
- deepmd/pt/train/training.py
- source/tests/pt/model/models/se_e2_a.json
🧰 Additional context used
🧠 Learnings (3)
📓 Common learnings
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
📚 Learning: 2025-12-12T13:40:14.325Z
Learnt from: CR
Repo: deepmodeling/deepmd-kit PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-12T13:40:14.325Z
Learning: Use PyTorch backend with `--pt` flag or `export DP_BACKEND=pytorch`; use `input_torch.json` for configuration typically
Applied to files:
source/tests/pt/test_init_frz_model.py
📚 Learning: 2025-12-12T13:40:14.325Z
Learnt from: CR
Repo: deepmodeling/deepmd-kit PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-12T13:40:14.325Z
Learning: Validate Python interface by running `python -c "import deepmd; import deepmd.tf; print('Both interfaces work')"`
Applied to files:
source/tests/pt/test_init_frz_model.py
🧬 Code graph analysis (1)
source/tests/pt/test_init_frz_model.py (1)
deepmd/entrypoints/convert_backend.py (1)
convert_backend(11-31)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (28)
- GitHub Check: Build wheels for cp311-macosx_x86_64
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Test C++ (true, false, false, true)
- GitHub Check: Build wheels for cp311-macosx_arm64
- GitHub Check: Build wheels for cp311-win_amd64
- GitHub Check: Test C++ (false, false, false, true)
- GitHub Check: Test C++ (true, true, true, false)
- GitHub Check: Test C++ (false, true, true, false)
- GitHub Check: Analyze (python)
- GitHub Check: Analyze (c-cpp)
- GitHub Check: Test Python (5, 3.10)
- GitHub Check: Test Python (3, 3.13)
- GitHub Check: Test Python (1, 3.10)
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
- GitHub Check: Test Python (6, 3.10)
- GitHub Check: Test Python (4, 3.10)
- GitHub Check: Test Python (4, 3.13)
- GitHub Check: Test Python (2, 3.10)
- GitHub Check: Test Python (5, 3.13)
- GitHub Check: Test Python (3, 3.10)
- GitHub Check: Test Python (6, 3.13)
- GitHub Check: Test Python (2, 3.13)
- GitHub Check: Test Python (1, 3.13)
- GitHub Check: Build C++ (clang, clang)
- GitHub Check: Build C++ (cuda120, cuda)
- GitHub Check: Build C++ (cpu, cpu)
- GitHub Check: Build C++ (rocm, rocm)
🔇 Additional comments (3)
source/tests/pt/test_init_frz_model.py (3)
16-18: LGTM! Imports support the new TF-to-PT conversion test.The added imports for
convert_backendandconvert_pbtxt_to_pbare necessary for the new test method and follow the existing import patterns in the file.Also applies to: 26-28
38-38: LGTM! UTF-8 encoding ensures cross-platform compatibility.Explicitly specifying UTF-8 encoding for JSON file operations is a good practice that prevents potential encoding issues across different platforms.
Also applies to: 67-67
174-175: LGTM! Proper cleanup of test artifacts.The teardown correctly removes
.pbfiles generated during the TF-to-PT conversion test, preventing test pollution and maintaining a clean test environment.
|
Please resolve the conservation of coderabbitai, then the pr can be merged. |
Head branch was pushed to by a user without write access
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
source/tests/pt/test_init_frz_model.py (1)
138-174: LGTM: Test successfully validates TF→PT frozen model initialization.The test properly validates the end-to-end conversion and initialization flow:
- Converts TF model (pbtxt→pb→pth) using the appropriate conversion utilities
- Initializes trainer from the converted frozen model
- Includes explicit assertions on trainer and model attributes (addressing previous feedback)
- Verifies the initialization completes without crashes
This directly tests the PR's objective of supporting
--init-frz-modelfor PT models converted from TF models.Optional enhancement: Consider adding inference verification to confirm the loaded model produces expected results:
# After line 173, optionally add: # Verify the model can perform inference dp = DeepPot(frozen_model) # Use test data from existing test_dp_test method # Run inference and verify resultsThis would provide stronger validation that the strict=False loading preserved model functionality, but is not essential for a basic smoke test.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
source/tests/pt/test_init_frz_model.py(5 hunks)
🧰 Additional context used
🧠 Learnings (4)
📓 Common learnings
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
📚 Learning: 2024-11-25T07:42:55.735Z
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
Applied to files:
source/tests/pt/test_init_frz_model.py
📚 Learning: 2025-12-12T13:40:14.325Z
Learnt from: CR
Repo: deepmodeling/deepmd-kit PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-12T13:40:14.325Z
Learning: Use PyTorch backend with `--pt` flag or `export DP_BACKEND=pytorch`; use `input_torch.json` for configuration typically
Applied to files:
source/tests/pt/test_init_frz_model.py
📚 Learning: 2025-12-12T13:40:14.325Z
Learnt from: CR
Repo: deepmodeling/deepmd-kit PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-12T13:40:14.325Z
Learning: Validate Python interface by running `python -c "import deepmd; import deepmd.tf; print('Both interfaces work')"`
Applied to files:
source/tests/pt/test_init_frz_model.py
🧬 Code graph analysis (1)
source/tests/pt/test_init_frz_model.py (1)
deepmd/entrypoints/convert_backend.py (1)
convert_backend(11-31)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (28)
- GitHub Check: Test Python (4, 3.10)
- GitHub Check: Test Python (6, 3.10)
- GitHub Check: Test Python (1, 3.10)
- GitHub Check: Test Python (6, 3.13)
- GitHub Check: Test Python (2, 3.10)
- GitHub Check: Test Python (5, 3.13)
- GitHub Check: Test Python (2, 3.13)
- GitHub Check: Test Python (4, 3.13)
- GitHub Check: Test Python (5, 3.10)
- GitHub Check: Test Python (3, 3.13)
- GitHub Check: Test Python (1, 3.13)
- GitHub Check: Test Python (3, 3.10)
- GitHub Check: Test C++ (true, true, true, false)
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
- GitHub Check: Build C++ (clang, clang)
- GitHub Check: Test C++ (false, false, false, true)
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Build wheels for cp311-macosx_arm64
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Test C++ (true, false, false, true)
- GitHub Check: Test C++ (false, true, true, false)
- GitHub Check: Build wheels for cp311-macosx_x86_64
- GitHub Check: Build wheels for cp311-win_amd64
- GitHub Check: Build C++ (cpu, cpu)
- GitHub Check: Build C++ (cuda120, cuda)
- GitHub Check: Build C++ (rocm, rocm)
- GitHub Check: Analyze (python)
- GitHub Check: Analyze (c-cpp)
🔇 Additional comments (3)
source/tests/pt/test_init_frz_model.py (3)
16-18: LGTM: Import additions support the new test.The new imports for
convert_backendandconvert_pbtxt_to_pbare necessary for the TF→PT model conversion test and are properly used.Also applies to: 26-28
38-38: LGTM: Explicit UTF-8 encoding is good practice.Making encoding explicit prevents platform-specific issues and is consistent with the new test method at line 149.
Also applies to: 67-67
185-186: LGTM: Cleanup handles new .pb artifacts.The tearDown method correctly removes the intermediate
.pbfile created during the pbtxt→pb→pth conversion, ensuring a clean test environment.
ChiahsinChu
force-pushed
the
devel-init-from-tf2pt
branch
from
2ddeeeb to
bb83aa9
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
source/tests/pt/test_init_frz_model.py (1)
66-71: Optional: simplify temp JSON handling and avoid leaving stray filesThe added UTF‑8 encoding is fine, but this pattern leaves a temp JSON file behind and reopens the
NamedTemporaryFileby name.You could simplify and self‑clean with something like:
- tmp_input = tempfile.NamedTemporaryFile(delete=False, suffix=".json") - with open(tmp_input.name, "w", encoding="utf-8") as f: - json.dump(empty_config, f, indent=4) - run_dp( - f"dp --pt train {tmp_input.name} --init-frz-model {self.models[-1]} --use-pretrain-script --skip-neighbor-stat" - ) + with tempfile.NamedTemporaryFile( + "w", delete=False, suffix=".json", encoding="utf-8" + ) as tmp_input: + json.dump(empty_config, tmp_input, indent=4) + tmp_name = tmp_input.name + run_dp( + f"dp --pt train {tmp_name} --init-frz-model {self.models[-1]} --use-pretrain-script --skip-neighbor-stat" + ) + os.remove(tmp_name)Not required for correctness, but it keeps
/tmp(or the platform temp dir) tidy and avoids double‑opening the same file.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
deepmd/pt/train/training.py(1 hunks)source/tests/pt/model/models/se_e2_a.json(1 hunks)source/tests/pt/test_init_frz_model.py(5 hunks)
🚧 Files skipped from review as they are similar to previous changes (2)
- source/tests/pt/model/models/se_e2_a.json
- deepmd/pt/train/training.py
🧰 Additional context used
🧠 Learnings (6)
📓 Common learnings
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
📚 Learning: 2024-11-25T07:42:55.735Z
Learnt from: njzjz
Repo: deepmodeling/deepmd-kit PR: 4302
File: deepmd/pd/infer/inference.py:35-38
Timestamp: 2024-11-25T07:42:55.735Z
Learning: In the file `deepmd/pd/infer/inference.py`, when loading the model checkpoint in the `Tester` class, it's acceptable to not include additional error handling for loading the model state dictionary.
Applied to files:
source/tests/pt/test_init_frz_model.py
📚 Learning: 2025-12-12T13:40:14.325Z
Learnt from: CR
Repo: deepmodeling/deepmd-kit PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-12T13:40:14.325Z
Learning: Run core tests with `pytest source/tests/tf/test_dp_test.py::TestDPTestEner::test_1frame -v` to validate basic functionality
Applied to files:
source/tests/pt/test_init_frz_model.py
📚 Learning: 2025-12-12T13:40:14.325Z
Learnt from: CR
Repo: deepmodeling/deepmd-kit PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-12T13:40:14.325Z
Learning: Test PyTorch training with `cd examples/water/se_e2_a && dp --pt train input_torch.json --skip-neighbor-stat` and allow timeout of 60+ seconds for validation
Applied to files:
source/tests/pt/test_init_frz_model.py
📚 Learning: 2025-12-12T13:40:14.325Z
Learnt from: CR
Repo: deepmodeling/deepmd-kit PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-12T13:40:14.325Z
Learning: Use PyTorch backend with `--pt` flag or `export DP_BACKEND=pytorch`; use `input_torch.json` for configuration typically
Applied to files:
source/tests/pt/test_init_frz_model.py
📚 Learning: 2025-12-12T13:40:14.325Z
Learnt from: CR
Repo: deepmodeling/deepmd-kit PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-12T13:40:14.325Z
Learning: Validate Python interface by running `python -c "import deepmd; import deepmd.tf; print('Both interfaces work')"`
Applied to files:
source/tests/pt/test_init_frz_model.py
🧬 Code graph analysis (1)
source/tests/pt/test_init_frz_model.py (2)
deepmd/entrypoints/convert_backend.py (1)
convert_backend(11-31)deepmd/pt/entrypoints/main.py (2)
main(519-578)get_trainer(96-204)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (28)
- GitHub Check: Build C++ (cuda120, cuda)
- GitHub Check: Test Python (5, 3.13)
- GitHub Check: Test Python (2, 3.13)
- GitHub Check: Test Python (6, 3.10)
- GitHub Check: Test Python (6, 3.13)
- GitHub Check: Build C++ (rocm, rocm)
- GitHub Check: Test Python (2, 3.10)
- GitHub Check: Test Python (3, 3.10)
- GitHub Check: Test Python (1, 3.10)
- GitHub Check: Build C++ (cpu, cpu)
- GitHub Check: Test Python (3, 3.13)
- GitHub Check: Test Python (4, 3.13)
- GitHub Check: Test Python (1, 3.13)
- GitHub Check: Build C++ (clang, clang)
- GitHub Check: Test Python (4, 3.10)
- GitHub Check: Test Python (5, 3.10)
- GitHub Check: Test C++ (false, true, true, false)
- GitHub Check: Test C++ (false, false, false, true)
- GitHub Check: Test C++ (true, false, false, true)
- GitHub Check: Test C++ (true, true, true, false)
- GitHub Check: Analyze (c-cpp)
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
- GitHub Check: Analyze (python)
- GitHub Check: Build wheels for cp311-macosx_arm64
- GitHub Check: Build wheels for cp311-manylinux_x86_64
- GitHub Check: Build wheels for cp311-macosx_x86_64
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Build wheels for cp311-win_amd64
🔇 Additional comments (4)
source/tests/pt/test_init_frz_model.py (4)
16-18: Conversion utility imports are well‑scoped and used
convert_backendandconvert_pbtxt_to_pbare imported only where needed and keep the test focused on the pbtxt→pb→pth path; no issues here.Also applies to: 26-28
38-39: Explicit UTF‑8 when reading JSON is a good robustness improvementOpening
se_atten.jsonwithencoding="utf-8"makes the test more portable across environments with different default encodings; the change is correct.
138-174: New pbtxt→pb→pth initialization test is well‑structuredThe new
test_init_frz_model_pb2pthcleanly exercises:
- conversion from
.pbtxtto.pb, then to.pth,get_trainer(..., init_frz_model=...)withnumb_steps = 0,- explicit assertions on
trainer,trainer.model, andtrainer.optimizer,- and a
trainer.run()call to ensure the init path doesn’t crash.This is a solid, clear end‑to‑end test for the new behavior.
175-186: Extended teardown correctly cleans new.pbartifactsAdding removal of
frozen_model*.pbaligns teardown with the new conversion test and keeps the working directory clean after runs; logic is consistent with the existing.pth/.ptcleanup.
tf2pt model) - Add try-catch block for loading frozen models - Use strict=False when state_dict keys don't match - Log warnings for model state_dict mismatches - Prevent crashes when loading models with different architectures"
ChiahsinChu
force-pushed
the
devel-init-from-tf2pt
branch
from
bb83aa9 to
e861add
Compare
…el (#5091) - Add try-catch block for loading frozen models - Use strict=False when state_dict keys don't match - Log warnings for model state_dict mismatches - Prevent crashes when loading models with different architectures" Fix bug in #5090 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Frozen model loading now uses non‑strict mode; missing or unexpected parameters are logged as warnings instead of causing failure. * **Tests** * Added an end-to-end test covering conversion and loading of frozen models from external formats. * Test I/O standardized to UTF‑8; teardown extended to remove generated artifacts. * **New Files** * Added a JSON test model configuration for validation and training workflows. <sub>✏️ Tip: You can customize this high-level summary in your review settings.</sub> <!-- end of auto-generated comment: release notes by coderabbit.ai -->
github-merge-queue
bot
removed this pull request from the merge queue due to no response for status checks
…el (deepmodeling#5091) - Add try-catch block for loading frozen models - Use strict=False when state_dict keys don't match - Log warnings for model state_dict mismatches - Prevent crashes when loading models with different architectures" Fix bug in deepmodeling#5090 <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Bug Fixes** * Frozen model loading now uses non‑strict mode; missing or unexpected parameters are logged as warnings instead of causing failure. * **Tests** * Added an end-to-end test covering conversion and loading of frozen models from external formats. * Test I/O standardized to UTF‑8; teardown extended to remove generated artifacts. * **New Files** * Added a JSON test model configuration for validation and training workflows. <sub>✏️ Tip: You can customize this high-level summary in your review settings.</sub> <!-- end of auto-generated comment: release notes by coderabbit.ai -->
4 participants
Fix bug in #5090
Summary by CodeRabbit
Bug Fixes
Tests
New Files
✏️ Tip: You can customize this high-level summary in your review settings.