# max_steps, the evaluator will send the final export signal. There is a
# small chance that the Estimator.train stopping logic sees a different
# global_step value (due to global step race condition and the fact the
- # saver sees a larger value for checkpoing saving), which does not end
+ # saver sees a larger value for checkpoint saving), which does not end
# the training. When the training ends, a new checkpoint is generated, which
# triggers the listener again. So, it could be the case the final export is
# triggered twice.
# TODO(akshayka): InputLayer should be a subclass of Layer, and it
# should implement the logic in input_layer using Layer's build-and-call
# paradigm; input_layer should create an instance of InputLayer and
-# return the result of inovking its apply method, just as functional layers do.
+# return the result of invoking its apply method, just as functional layers do.
class InputLayer(object):
"""An object-oriented version of `input_layer` that reuses variables."""
tensor_name_in_ckpt=None, max_norm=None, trainable=True):
"""List of dense columns that convert from sparse, categorical input.
- This is similar to `embedding_column`, except that that it produces a list of
+ This is similar to `embedding_column`, except that it produces a list of
embedding columns that share the same embedding weights.
Use this when your inputs are sparse and of the same type (e.g. watched and