extending the mlp module

[wyli]

Signed-off-by: Wenqi Li wenqil@nvidia.com

Description

(this is a follow-up of PR #4074 and the discussions.)

This PR extends the MLPBlock (multi-layer perceptron) with

• configurable activation function
• shared or independent dropout instances

the refactoring is inspired by #4074 #4074 (comment) and the goal is to

• enhance the flexibility of the essential building blocks and
• encourage the reuse of the existing components when bringing in new network models.

previously trained models (such as MMAR) can still be used with this update
verified: https://github.com/Project-MONAI/MONAI/runs/5881423584?check_suite_focus=true

Status

Ready

Types of changes

• Non-breaking change (fix or new feature that would not break existing functionality).
• Breaking change (fix or new feature that would cause existing functionality to change).
• New tests added to cover the changes.
• Integration tests passed locally by running ./runtests.sh -f -u --net --coverage.
• Quick tests passed locally by running ./runtests.sh --quick --unittests --disttests.
• In-line docstrings updated.
• Documentation updated, tested make html command in the docs/ folder.

[wyli]

this is compatible with previously released model weights, verified here https://github.com/Project-MONAI/MONAI/runs/5881423584?check_suite_focus=true

[wyli]

/build

[ericspod]

I would add some description or comment explaining why this is done and what the meaning is of the modes.

[ahatamiz]

Looks good. Please merge so we can update the swin unetr PR.

[Nic-Ma]

Hi @wyli ,

I think @yiheng-wang-nv @ahatamiz approved the PR, do you have any missing parts to add or update?

Thanks.

[wyli]

/build