add VideoSwin

[divyashreepathihalli]

This task is part of the KerasHub project ( Consolidating all of KerasCV and KerasNLP models in to one place)
Models from KerasCV will be added to KerasNLP in KerasNLP infrstructure style. The development is now under the keras-hub branch in KerasNLP - https://github.com/keras-team/keras-nlp/tree/keras-hub

Steps to complete this Task

• Source code : https://github.com/keras-team/keras-cv/tree/master/keras_cv/src/models/backbones/video_swin
• Destination : A model folder here https://github.com/keras-team/keras-nlp/tree/keras-hub/keras_nlp/src/models
• Example PR : Add VGG16 and VGG19 backbone #1737, please follow the style from here and take a look at the comments on the PR for any design clarifications
• copy the model over
• If the model has multiple versions with small changes in architecture, they should be added as one model and the configuration should be handled in the args
• use KerasNLP test routines to update the tests
• for reformatting the code run shell/api_gen.sh, shell/format.sh and shell/lint.sh
• The backbone will be added as a model, any downstream task model will be under the same model folder - example the video classifier task model.

Add either @divyashreepathihalli, @mattdangerw or @SamanehSaadat for review of any PR.

[kernel-loophole]

@divyashreepathihalli I'm interested in contributing to this task for the KerasHub project by adding models from KerasCV to KerasNLP in the specified format. Before I get started, could you clarify a few things?

File Locations: Could you confirm if the full path for the source files is keras-cv/keras_cv/src/models/backbones/video_swin? Additionally, should all files within this folder be moved to a new folder under keras-nlp/keras_nlp/src/models?

Model Configuration: For models with multiple versions, do you have any preferred method or example in KerasNLP to manage these configurations within the model’s arguments?

Testing Requirements: Are there any specific test cases or benchmarks for video models that need to be prioritized during testing, or should I follow the general test routines provided in KerasNLP?

[kernel-loophole]

@divyashreepathihalli is this model will be placed in this path https://github.com/keras-team/keras-hub/tree/keras-hub/keras_nlp/src/models and how can test if the model is working or not

[Swap98]

@divyashreepathihalli I'm interesting in contributing to this task of adding models from KerasCV to KerasNLP

[divyashreepathihalli]

Hey @Swap98 looks like @kernel-loophole responded on the PR. He might be actively looking into it.