[FEA] Create unit tests for saving and loading back the retrieval and ranking models
🚀 Feature request
we are missing unit tests for saving and loading back the trained models. This is important for inference step and to run inference examples without issues. We should be sure that we can save and load back the models without issues, so that, unit tests would help for that.
We have a unit test that can be utilized for this purpose as well: https://github.com/NVIDIA-Merlin/models/blob/main/merlin/models/tf/utils/testing_utils.py#L73
waiting for https://github.com/NVIDIA-Merlin/models/issues/718 for retrieval.
For ranking, if we have a BC case, we can do that with model.predict(). once there was a PR with an example for generating predictions from a DLRM model: https://github.com/NVIDIA-Merlin/models/blob/getting_started_exp/examples/getting-started/ranking/DLRM_with_prediction.ipynb