Use config which has been previously validated

[vreyespue]

The input param config might be a str or a dict, in case of string the command config.get will generate an error. Using the already validated self.config should avoid a potential error.

Code Pull Requests

Please provide the following:

• a clear explanation of what your code does
• if applicable, a reference to an issue
• a reproducible test for your PR (code, config and data sample)

Documentation Pull Requests

Note that the documentation HTML files are in docs/ while the Markdown sources are in mkdocs/docs.

If you are proposing a modification to the documentation you should change only the Markdown files.

api.md is automatically generated from the docstrings in the code, so if you want to change something in that file, first modify ludwig/api.py docstring, then run mkdocs/code_docs_autogen.py, which will create mkdocs/docs/api.md .

[w4nderlust]

This is a great catch! Thank you

[vreyespue]

@w4nderlust just a small question, 2 integration tests did not pass, specifically:

=========================== short test summary info ============================
FAILED tests/integration_tests/test_regularizers.py::test_encoder[test_case3]
FAILED tests/integration_tests/test_regularizers.py::test_encoder[test_case6]
= 2 failed, 1092 passed, 16 skipped, 78 deselected, 1915 warnings in 1280.21s (0:21:20) =
Error: Process completed with exit code 1.

However, they are green when I run those tests locally including the change proposed in this PR:

~/ludwig-ai/ludwig git:(master) pytest tests/integration_tests/test_regularizers.py
=== 10 passed, 7 warnings in 3.52s ===

Therefore I guess that these 2 test failures are unrelated to the fix proposed in this PR.
What do you think? Should I trigger the execution of the tests again?
Many thanks and best regards.

[w4nderlust]

Yes don't worry, it's an issue with tfnightly that we will solve in another PR. Thank you for you contribution, merging it!

[vreyespue]

Ok cool, many thanks 👍