diff options
| author |  Mustafa Ispir <ispir@google.com> | 2018-08-10 10:56:50 -0700 |
|---|---|---|
| committer |  TensorFlower Gardener <gardener@tensorflow.org> | 2018-08-10 11:01:06 -0700 |
| commit | 729caa48a34683cd38bb14c48bb63e8cddc88d60 (patch) | |
| tree | 90a567ff5e565b36afcc335e16d96e4ae56b8b2e | |
| parent | cae772e4436dce73e82256150029a9d250b800a1 (diff) | |
Update the docs to clear re-creation of evaluation graph.
Fixes #19062
PiperOrigin-RevId: 208235214
| -rw-r--r-- | tensorflow/python/estimator/training.py | 4 |
1 files changed, 4 insertions, 0 deletions
diff --git a/tensorflow/python/estimator/training.py b/tensorflow/python/estimator/training.py index a01b2300dd..bb1305767f 100644 --- a/tensorflow/python/estimator/training.py +++ b/tensorflow/python/estimator/training.py @@ -323,6 +323,10 @@ def train_and_evaluate(estimator, train_spec, eval_spec): tf.estimator.train_and_evaluate(estimator, train_spec, eval_spec) ``` + Note that in current implementation `estimator.evaluate` will be called + multiple times. This means that evaluation graph (including eval_input_fn) + will be re-created for each `evaluate` call. `estimator.train` will be called + only once. Example of distributed training: |