Common & Submission Configuration#
- pydantic model dpeva.config.BaseWorkflowConfig[源代码]#
Base configuration for all workflows.
- field dp_backend: str = 'pt'#
DeepMD-kit backend (e.g. 'pt', 'tf', 'jax', 'pd').
- field omp_threads: int | Literal['auto'] = 4#
OpenMP threads (set to 'auto' to use all cores).
- field submission: SubmissionConfig [Optional]#
Submission configuration.
- field work_dir: Path [Optional]#
Working directory.
- validator validate_dp_backend » dp_backend[源代码]#
Validates that the DeepMD backend is supported.
- validator validate_omp_threads » omp_threads[源代码]#
Validates OMP thread count and resolves 'auto'.
- pydantic model dpeva.config.SubmissionConfig[源代码]#
Configuration for job submission.
- field backend: Literal['local', 'slurm'] = 'local'#
Execution backend.
- field env_setup: str | List[str] = ''#
List of shell commands to execute before running the task (e.g., module load cuda).
- field slurm_array: bool = False#
When true, workflows that support homogeneous Slurm arrays may submit arrays instead of one job per task.
- field slurm_array_task_limit: int | None = None#
Optional Slurm array throttle, rendered as --array=0-N%limit.
- field slurm_config: Dict[str, Any] [Optional]#
Slurm specific parameters (partition, nodes, etc).