Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Custom executor

See original GitHub issue

According to https://www.tensorflow.org/tfx/guide/examplegen it’s possible to use a custom Executor in this way:

from tfx.components.example_gen.csv_example_gen import executor
from tfx.utils.dsl_utils import external_input

examples = external_input(os.path.join(base_dir, 'data/simple'))
example_gen = FileBasedExampleGen(input_base=examples,
                                  executor_class=executor.Executor)

I’m confused about that, FileBasedExampleGen doesn’t accept executor_class as a parameter. Maybe I should use custom_executor_spec?

I’m also not sure about how to import this class into the pipeline itself. AS far as I know ExampleGen components lack the module_file property that Transform and Trainer have.

Am I missing smething?

Issue Analytics

State:
Created 4 years ago
Comments:7 (4 by maintainers)

Top GitHub Comments

2reactions

hongye-suncommented, Sep 5, 2019

@valeriano-manassero, executor_class was changed to custom_executor_spec in https://github.com/tensorflow/tfx/commit/3069981910a5011ddac86a01ad90a11d637ede5b. We will update the doc to reflect the change soon.

Instead of writing executor_class=executor.Executor, you need to write custom_executor_spec=executor_spec.ExecutorClassSpec(executor.Executor).

Importing executor class is like importing other python module or class, as long as the python file is under path in sys.path list, you should be able to import it in your pipeline.

0reactions

valeriano-manasserocommented, Sep 6, 2019

@hongye-sun it makes perfectly sense, ty for info. I’m going to create a build to achieve the goal.