deepmd.jax.entrypoints.train

DeePMD training entrypoint script.

Can handle local training.

Functions

train(→ None)

Run DeePMD model training.

Module Contents

deepmd.jax.entrypoints.train.train(*, INPUT: str, init_model: str | None, restart: str | None, output: str, init_frz_model: str | None, mpi_log: str, log_level: int, log_path: str | None, skip_neighbor_stat: bool = False, finetune: str | None = None, use_pretrain_script: bool = False, **kwargs: Any) → None

Run DeePMD model training.

Parameters:

INPUTstr

json/yaml control file

init_modelOptional[str]

path prefix of checkpoint files or None

restartOptional[str]

path prefix of checkpoint files or None

outputstr

path for dump file with arguments

init_frz_modelstr | None

path to frozen model, or None if no frozen model is used

mpi_logstr

mpi logging mode

log_levelint

logging level defined by int 0-3

log_pathOptional[str]

logging file path or None if logs are to be output only to stdout

skip_neighbor_statbool, default=False

skip checking neighbor statistics

finetuneOptional[str]

path to pretrained model or None

use_pretrain_scriptbool

Whether to use model script in pretrained model when doing init-model or init-frz-model. Note that this option is true and unchangeable for fine-tuning.

**kwargs

additional arguments

Raises:

RuntimeError

if the training command fails.