Properly handle parent modules w/ parameters in BaseFinetuning callback

[speediedan]

What does this PR do?

Fixes #7930

This PR fixes issue #7930 and provides relevant test coverage: test_finetuning_callback.py::test_parent_module_w_param_model

Models w/ parent modules that directly contained parameters themselves (as opposed to parameters exclusively in their constituent submodules) had their parameters ignored by the BaseFinetuning callback. The primary issue was addressed with an additional filtering condition in BaseFinetuning.flatten_modules(). To avoid passing duplicate parameters in methods that depended upon BaseFinetuning.flatten_modules, I also added recurse=False to the parameter loops in BaseFinetuning.[freeze, make_trainable, filter_params]. The new test I've added could potentially be consolidated with /tests/callbacks/test_finetuning_callback.py::test_deep_nested_model but I at least initially have broken the test out separately for clarity. As a practical example of the issue, I initially encountered this bug when using BaseFinetuning w/ deberta, specifically, the DisentangledSelfAttention parent module

I've updated the docstring of flatten_modules() to reflect the changes. All test_finetuning_callback.py tests are passing on my end.

• Was this discussed/approved via a GitHub issue? (not for typos and docs)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• Did you make sure to update the documentation with your changes? (if necessary)
• Did you write any new necessary tests? (not for typos and docs)
• Did you verify new and existing tests pass locally with your changes?
• Did you update the CHANGELOG? (not for typos, docs, test updates, or internal minor changes/refactorings)

PR review

Anyone in the community is free to review the PR once the tests have passed.
Before you start reviewing make sure you have read Review guidelines. In short, see the following bullet-list:

• Is this pull request ready for review? (if not, please submit in draft mode)
• Check that all items from Before submitting are resolved
• Make sure the title is self-explanatory and the description concisely explains the PR
• Add labels and milestones (and optionally projects) to the PR so it can be classified

Did you have fun?

Make sure you had fun coding 🙃
YES! This is my first contribution to PL but I'm hoping to be more involved in the future. Thanks for the awesome product/work!

[codecov]

Codecov Report

Merging #7931 (2350723) into master (22d8266) will decrease coverage by 5%.
The diff coverage is 100%.

@@           Coverage Diff           @@
##           master   #7931    +/-   ##
=======================================
- Coverage      92%     87%    -5%     
=======================================
  Files         203     203            
  Lines       13127   13127            
=======================================
- Hits        12013   11395   -618     
- Misses       1114    1732   +618     

[carmocca]

Nice catch!

Remember to update the CHANGELOG.md file

[tchaton]

Good catch ! Thanks for using the BaseFinetuning Callback and catching this bug.