Fix an API inconsistency (system_config is a named parameter in orb) #12
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Introduce an abstract property_type method to PropertyConfigBase and implement it in subclasses. This enhances the type system by explicitly defining whether a property is system-wide or atom-specific.
Simplify constraint detection logic and handle multiple FixAtoms constraints. This change enhances robustness and flexibility when processing atom constraints in the _get_fixed function.
Adds explicit assertions to clarify that conservative forces and stresses are not currently supported in the JMP model yet. This improves error handling and provides clearer feedback to users attempting to use unsupported configurations.
Introduces a new 'reduction' field to GraphPropertyConfig, allowing customization of how property values are aggregated across atoms. This enhances flexibility for handling different types of properties (extensive, intensive, or specialized cases like phonon peaks).
Implements a new output head for graph-level scalar predictions in the JMP backbone. This enhances the model's capability to handle graph property tasks with customizable reduction methods.
Improve flexibility by making 'properties' parameter optional. Update type hints for better compatibility with ASE. Simplify property resolution logic for cleaner code.
Adds detailed docstrings to key methods in ASE calculator and potential wrappers, clarifying purpose, parameters, return values, and internal logic. Improves code readability and assists developers in understanding the prediction process for atomic systems.
Improve code documentation by adding a detailed docstring to the MatterTunePotential class. The new docstring explains the class's purpose, its parameters, and provides usage examples, enhancing code clarity and usability.
Extracts the Voigt notation to 3x3 stress matrix conversion function to a shared utility file, promoting code reuse and maintainability.
- Implement EqV2BackboneConfig and EqV2BackboneModule - Create output heads for energy, forces, and stresses, as well as arbitrary scalar graph properties - Include example notebook demonstrating usage for OMAT24 dataset
Eqv2 support
Improve error messages for missing dependencies, specifically for orb_models. Add note about temporary patched version requirement and link to open PR for device argument fix. Also, added a TODO for loading graph generation config from checkpoint for equiformer (although we should look into doing this for other models too)
I was originally under the impression that eqv2's 6 dim stress was Voigt notation, but the methodology seems to be quite similar (based off the same base codebase) as the GemNet-OC stress calculations. I have updated the code to reflect this.
The return type annotation for the abstract hparams_cls method is changed from FinetuneModuleBaseConfig to TFinetuneModuleConfig.
Enhances README with complete workflow example showing: - Model fine-tuning phase - Practical usage with ASE integration - Geometry optimization example Updates license section with direct link to documentation
Feature: load dataset from a custom JSON file
…xport order for the config exports, preventing unnecessary condegen changesets in the future
Removes unused Python imports and package references to reduce code clutter and improve maintainability. Updates configuration files to enable F401 lint rule for detecting unused imports in future.
Previously, we used Poetry to manage our dependencies and build process. This doesn't work well with PyTorch as it has some difficult-to-install dependencies. For now, I've just changed this to be a standard setuptools-based pyproject.toml file. We can revisit this in the future once we have a better solution.
Simplifies pre-commit configuration by: - Updating ruff-pre-commit to v0.8.1 - Removing redundant ruff hook - Changing --select to --extend-select for imports - Removing local pyright hook
Adds codegen'd files to .gitattributes file.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.