Uber Ludwig Toolbox: export generated model to Google platform ML Engine

Question

In Uber Ludwig toolbox the training from the scratch produced the following 10 files:

1. checkpoint
2. model_hyperparameters.json
3. model_weights.data-00000-of-00001
4. model_weights.index
5. model_weights.meta
6. model_weights_progress.data-00000-of-00001
7. model_weights_progress.index
8. model_weights_progress.meta
9. train_set_metadata.json
10. training_progress.p

I would like to export them (or only the needed ones) into one file saved_model.pb to be able to load it to my ML Engine Google platform.

I tried the freeze_graph.py script, but it requires as an input a .pb file which I do not have. (I have only the above-mentioned 10 files). How to proceed to get this one saved_model.pb file? I saw several threads, but none of them works for me.

Answer 1

Ludwig added support for SavedModel which is the only model format supported by AI Platform (Cloud ML Engine)

Take a look at this issue