experiments
/
STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Add a note to `model_to_estimator` docs about weighted metrics. 

PiperOrigin-RevId: 307971602
Change-Id: Ia4c5525b0bb8fbcb6f56b32fc5ad27faaf774568
This commit is contained in:
Pavithra Vijay 2020-04-22 21:36:22 -07:00 committed by TensorFlower Gardener
parent ced65fa9b0
commit 6889b4f91c
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
tensorflow/python/keras/estimator/__init__.py
View File

@ -169,6 +169,11 @@ def model_to_estimator_v2(keras_model=None,
estimator.train(input_fn, steps=1)
```
Note: We do not support creating weighted metrics in Keras and converting them
to weighted metrics in the Estimator API using `model_to_estimator`.
You will have to create these metrics directly on the estimator spec using the
`add_metrics` function.
To customize the estimator `eval_metric_ops` names, you can pass in the
`metric_names_map` dictionary mapping the keras model output metric names
to the custom names as follows: