Need better documentation for TFMA's MetricConfig for TFX's Evaluator

[sumitbinnani]

URL(s) with the issue:

• https://www.tensorflow.org/tfx/model_analysis/api_docs/python/tfma/MetricConfig

• https://github.com/tensorflow/model-analysis/blob/4ea7d16125d26aa469dc2c5199bbb5f588586fef/tensorflow_model_analysis/proto/config.proto#L407-L414

Description of issue (what needs changing):

For TFX's Evaluator, it seems like the config does not rely on the __init__ method; and relies on the get_config and from_config method of the (custom) metrics class.

Clear description:

It seems like the config does not rely on the __init__ method; and relies on the get_config and from_config method of the (custom) metrics class. This distinction is important as TFX's Evaluator serialize and deserialize the metric and __init__ is not used during the final metrics evaluation.

Correct links

It seems like the config does not rely on the __init__ method; and relies on the get_config and from_config method of the (custom) metrics class. This distinction is important as TFX's Evaluator serialize and deserialize the metric and __init__ is not used during the final metrics evaluation.

Parameters defined

Yes. However, the definition of the config breaks for TFX Evaluator

Returns defined

Yes.

Raises listed and defined

No

Request visuals, if applicable

No

[sumitbinnani]

Link to TFMA Issue: tensorflow/model-analysis#134

[mdreves]

Closing, will track in tensorflow/model-analysis#134