deepmd.tf.train.run_options

Module taking care of important package constants.

Classes

RunOptions

Class with info on how to run training (cluster, MPI and GPU config).

Module Contents

class deepmd.tf.train.run_options.RunOptions(init_model: str | None = None, init_frz_model: str | None = None, finetune: str | None = None, restart: str | None = None, log_path: str | None = None, log_level: int = 0, mpi_log: str = 'master')

Class with info on how to run training (cluster, MPI and GPU config).

Attributes:

gpus: Optional[list[int]]

list of GPUs if any are present else None

is_chief: bool

in distribured training it is true for the main MPI process in serail it is always true

world_size: int

total worker count

my_rank: int

index of the MPI task

nodename: str

name of the node

node_list_list[str]

the list of nodes of the current mpirun

my_device: str

device type - gpu or cpu

gpus: list[int] | None

world_size: int

my_rank: int

nodename: str

nodelist: list[int]

my_device: str

_HVD: horovod.tensorflow | None

_log_handles_already_set: bool = False

restart = None

init_model = None

init_frz_model = None

finetune = None

init_mode = 'init_from_scratch'

property is_chief

Whether my rank is 0.

print_resource_summary() → None

Print build and current running cluster configuration summary.

_setup_logger(log_path: pathlib.Path | None, log_level: int, mpi_log: str | None) → None

Set up package loggers.

Parameters:

log_levelint

logging level

log_pathOptional[str]

path to log file, if None logs will be send only to console. If the parent directory does not exist it will be automatically created, by default None

mpi_logOptional[str], optional

mpi log type. Has three options. master will output logs to file and console only from rank==0. collect will write messages from all ranks to one file opened under rank==0 and to console. workers will open one log file for each worker designated by its rank, console behaviour is the same as for collect.

_try_init_distrib() → None

_init_distributed(HVD: RunOptions._init_distributed.HVD) → None

Initialize settings for distributed training.

Parameters:

HVDHVD

horovod object

_init_serial() → None

Initialize setting for serial training.