hook_checkpoint_saver: Saves Checkpoints Every N Steps or Seconds

In tfestimators: Interface to 'TensorFlow' Estimators

Saves Checkpoints Every N Steps or Seconds

Description

Saves Checkpoints Every N Steps or Seconds

Usage

hook_checkpoint_saver(
  checkpoint_dir,
  save_secs = NULL,
  save_steps = NULL,
  saver = NULL,
  checkpoint_basename = "model.ckpt",
  scaffold = NULL,
  listeners = NULL
)

Arguments

checkpoint_dir	The base directory for the checkpoint files.
save_secs	An integer, indicating saving checkpoints every N secs.
save_steps	An integer, indicating saving checkpoints every N steps.
saver	A saver object, used for saving.
checkpoint_basename	The base name for the checkpoint files.
scaffold	A scaffold, used to get saver object.
listeners	List of checkpoint saver listener subclass instances, used for callbacks that run immediately after the corresponding hook_checkpoint_saver callbacks, only in steps where ⁠the hook_checkpoint_saver⁠ was triggered.

See Also

Other session_run_hook wrappers: hook_global_step_waiter(), hook_history_saver(), hook_logging_tensor(), hook_nan_tensor(), hook_progress_bar(), hook_step_counter(), hook_stop_at_step(), hook_summary_saver(), session_run_hook()

tfestimators documentation built on Aug. 19, 2025, 1:15 a.m.