How do I load a keras saved model with custom Optimizer

Question

I have compiled and trained a keras model with a custom optimizer. I saved the model but when I try to load the model, it throws an error stating ValueError: Unknown optimizer: MyOptimizer. I tried to pass MyOptimizer as a custom object something like : models.load_model('myModel.h5', custom_objects={'optimizer':MyOptimizer}) and it still throws an error. How do I load the model a keras model with custom Objects?

To save/load state of a custom optimizer: https://stackoverflow.com/questions/49503748/save-and-load-model-optimizer-state — Contango, Mar 13 '23 at 20:59

score 18 · Answer 1 · answered Jun 12 '19 at 15:33

I ran into the same problem :)

I made it work by loading the model with models.load_model('myModel.h5', compile=False).

From the keras source code:

If an optimizer was found as part of the saved model, the model is already compiled. Otherwise, the model is uncompiled and a warning will be displayed. When compile is set to False, the compilation is omitted without any warning.

After the uncompiled model is loaded, I can compile it again with my custom optimizer.

btw, you don't need to compile for inference only – Eugene Krevenets Aug 01 '19 at 09:59 — Eugene Krevenets, Aug 01 '19 at 09:59

score 9 · Answer 2 · answered Dec 02 '19 at 07:13

9

You have to use the name of optimizer class as the key in the custom_objects dictionary, in your case, as the optimizer would be 'MyOptimizer' object,

models.load_model('myModel.h5', custom_objects={'MyOptimizer': MyOptimizer})

should work

answered Dec 02 '19 at 07:13

SajanGohil

960
13
26

score 1 · Answer 3 · answered Jan 01 '21 at 12:34

I had the same problem. However, I had two different custom things in my model. One was my optimizer and the other was a custom layer. Therefore, I solved my problem as follow:

my_loaded_model = tf.keras.models.load_model('my_models_name.h5', custom_objects={'KerasLayer':hub.KerasLayer , 'AdamWeightDecay': optimizer})

How do I load a keras saved model with custom Optimizer

3 Answers3

Linked