directly save encoder and decoder models, as well as weights

[quantumjot]

Here:

cellx-predict/cellxpredict/train.py

Line 57 in 9e43599

model.encoder.save_weights(model_filename.with_suffix(".h5"))

we should also be able to save a model (rather than just weights, since quantumjot/cellx#44 fixed the custom layer serialization). This would make loading and inference much easier...