Documentation

NPUEstimator Constructor

Description

Constructor of the NPUEstimator class. The NPUEstimator class inherits the Estimator class of TensorFlow and can call the native APIs of the base class to train and evaluate TensorFlow models.

Prototype

def __init__(self,
model_fn=None,
model_dir=None,
config=None,
params=None,
job_start_file='',
warm_start_from=None
)

Options

Option

Input/Output

Description

model_fn

Input

Model function definition. This function returns an object of the NPUEstimatorSpec class.

For details about the constructor of the NPUEstimatorSpec class, see NPUEstimatorSpec Constructor.

model_dir

Input

Model directory, which is used to save or restore model files. Defaults to None.

If model_dir set in NPURunConfig is different from that in NPUEstimator, an error is reported.

If either NPURunConfig or NPUEstimator is configured with model_dir, the configured path applies.

If neither NPURunConfig nor NPUEstimator is configured with model_dir, a model_dir_xxxxxxxxxx directory is created in the current script execution path to save the model file.

config

Input

Object of the NPURunConfig class.

For details about the constructor of the NPURunConfig class, see NPURunConfig Constructor.

params

Input

Argument of model_fn, which is of the dictionary type. The key is the name of the argument, and the value is the basic Python type value.

job_start_file

Input

Startup file path of the CSA job

warm_start_from

Input

Path of the checkpoint. The checkpoint will be imported for training.

Returns

An object of the NPUEstimator class

Examples

1
2
3
4
5
6
7
8
from npu_bridge.npu_init import *
...
self._classifier=NPUEstimator(
  model_fn=cnn_model_fn,
  model_dir=self._model_dir,
  config=tf.estimator.NPURunConfig(
      save_checkpoints_steps=50 if get_rank_id() == 0 else 0,
      keep_checkpoint_max=1))