Create Custom Session Run Hooks

Create a set of session run hooks, used to record information during training of an estimator. See Details for more information on the various hooks that can be defined.

session_run_hook(begin = function() { },
  after_create_session = function(session, coord) { },
  before_run = function(context) { }, after_run = function(context, values)
  { }, end = function(session) { })

Arguments

begin

function(): An R function, to be called once before using the session.

after_create_session

function(session, coord): An R function, to be called once the new TensorFlow session has been created.

before_run

function(run_context): An R function to be called before a run.

after_run

function(run_context, run_values): An R function to be called after a run.

end

function(session): An R function to be called at the end of the session.

Typically, you'll want to define a before_run() hook that defines the set of tensors you're interested in for a particular run, and then you'll use the resulting values of those tensors in your after_run() hook. The tensors requested in your before_run() hook will be made available as part of the second argument in the after_run() hook (the values argument).

See also

session_run_args()

Other session_run_hook wrappers: hook_checkpoint_saver, 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