chore: update dpa3 example #4778
Conversation
📝 WalkthroughWalkthroughA new dynamic training configuration file for the DPA3 model was added, specifying detailed model, fitting, loss, and training parameters. The existing static configuration was updated with modified descriptor parameters. The test suite was updated to include the new dynamic configuration file for validation. Changes
Sequence Diagram(s)sequenceDiagram
participant Tester as TestExamples
participant ConfigStatic as input_torch.json
participant ConfigDynamic as input_torch_dynamic.json
Tester->>ConfigStatic: Validate arguments
Tester->>ConfigDynamic: Validate arguments
Suggested reviewers
✨ Finishing Touches
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. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. 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)
Other keywords and placeholders
CodeRabbit Configuration File (
|
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
examples/water/dpa3/input_torch.json (1)
27-29: Document and validate new descriptor flagsThree new boolean parameters (
smooth_edge_update,edge_init_use_dist,use_exp_switch) were added. Please ensure:
- The codebase recognizes and applies these flags in the descriptor update routines.
- Documentation (e.g., README, config schema) is updated to explain their effects.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
examples/water/dpa3/input_torch.json(1 hunks)examples/water/dpa3/input_torch_dynamic.json(1 hunks)source/tests/common/test_examples.py(1 hunks)
⏰ Context from checks skipped due to timeout of 90000ms (29)
- GitHub Check: Build C library (2.14, >=2.5.0rc0,<2.15, libdeepmd_c_cu11.tar.gz)
- GitHub Check: Build C library (2.18, libdeepmd_c.tar.gz)
- 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: Build C++ (cpu, cpu)
- GitHub Check: Test C++ (true)
- GitHub Check: Test C++ (false)
- GitHub Check: Build wheels for cp311-win_amd64
- GitHub Check: Build wheels for cp310-manylinux_aarch64
- GitHub Check: Build wheels for cp311-macosx_x86_64
- 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: Test Python (6, 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 (3, 3.12)
- GitHub Check: Test Python (4, 3.12)
- GitHub Check: Test Python (3, 3.9)
- GitHub Check: Test Python (2, 3.12)
- GitHub Check: Test Python (1, 3.9)
- GitHub Check: Test Python (2, 3.9)
- GitHub Check: Test Python (4, 3.9)
- GitHub Check: Test Python (1, 3.12)
- GitHub Check: Analyze (python)
- GitHub Check: Analyze (c-cpp)
🔇 Additional comments (9)
examples/water/dpa3/input_torch.json (3)
11-12: Verify reduced feature dimensionsThe node (
n_dim) and edge (e_dim) feature dimensions have been halved. This affects model capacity, weight shapes, and memory usage. Please confirm that:
- All architecture definitions and weight initializers have been updated to match these new dimensions.
- Any downstream code or pre-trained checkpoints expecting the old sizes have been adapted accordingly.
16-19: Validate updated smoothing cutoff distancesThe smoothing cutoff radii (
e_rcut_smthfrom 3.0→5.3 anda_rcut_smthfrom 2.0→3.5) can significantly change neighbor inclusion and computational cost. Ensure that:
- The new cutoff distances align with the desired physical interaction ranges.
- Upstream code calculating neighbor lists or edge weights correctly uses these updated values.
22-22: Confirm replacement ofskip_statwithfix_stat_stdThe boolean
skip_statflag was removed and replaced byfix_stat_std: 0.3. Verify that:
- Configuration parsing logic no longer references
skip_stat.fix_stat_stdis documented and handled in the model builder with appropriate defaults.examples/water/dpa3/input_torch_dynamic.json (5)
11-35: Review dynamicrepflowdescriptor parametersThe dynamic configuration extends the static descriptor with:
- Selection sizes (
e_sel: 1200,a_sel: 300)- Dynamic selection controls (
use_dynamic_sel,sel_reduce_factor)- Other hyperparameters inherited from the static config
Please verify that:
- The model initialization code fully supports these new dynamic parameters.
- Reasonable defaults or fallbacks exist if any key is omitted.
- The expanded
e_sel/a_selvalues are compatible with memory constraints.
41-52: Approve fitting network configurationThe three-layer fitting network (
neuron: [240,240,240]), residual connections (resnet_dt: true), and float32 precision mirror the static config. Everything looks consistent.
55-61: Approve learning rate scheduleThe exponential decay schedule (
start_lr: 0.001→stop_lr: 3e-5overdecay_steps: 5000) aligns with existing conventions.
62-71: Approve loss function parametersThe energy‐based loss with start/limit prefactors for energy, forces, and virial terms is correctly specified.
72-99: Approve training data and hyperparametersThe training settings—including data paths, batch sizes, step counts, gradient clipping, and output frequencies—are consistent with project standards.
source/tests/common/test_examples.py (1)
61-63: Include dynamic config in example testsAdding
input_torch_dynamic.jsontoinput_filesensures the new dynamic DPA3 example is validated by the argument checker.
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## devel #4778 +/- ##
=======================================
Coverage 84.78% 84.78%
=======================================
Files 698 698
Lines 67755 67755
Branches 3542 3544 +2
=======================================
+ Hits 57448 57449 +1
+ Misses 9174 9173 -1
Partials 1133 1133 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Summary by CodeRabbit
New Features
Enhancements
Tests