Basically, we borrowed from core Keras' common layer testing routes here, and extended them to KerasNLP abstractions:
TestCase.run_layer_test()
TestCase.run_backbone_test()
TestCase.run_task_test()
This has proved to be quite useful for maintenance of the library. It makes it much easier to plug coverage holes across all models in a quick PR. We should consider doing the same for CV!
divyashreepathihalli
commented
7 months ago
A few months ago, we landed a large refactoring of KerasNLP testing routines: https://github.com/keras-team/keras-nlp/pull/1269
Basically, we borrowed from core Keras' common layer testing routes here, and extended them to KerasNLP abstractions:
TestCase.run_layer_test()TestCase.run_backbone_test()TestCase.run_task_test()This has proved to be quite useful for maintenance of the library. It makes it much easier to plug coverage holes across all models in a quick PR. We should consider doing the same for CV!