[FEATURE REQUEST] add support for custom layers in `best_model()`

Overview

I built a model in Keras using the functional API. I also use the keras_contrib and keras_radam libraries to add new activations (Swish) and optimizers (RAdam) not yet implemented in keras. Talos initializes and trains all iterations of the model without issue, but if I want to recall the best model or deploy the model, it fails with an error from keras.utils.generic_utils.deserialize_keras_object(). The error in question is ValueError: Unknown layer: Swish.

Prerequisites

• My Python version is 3.5 or higher
• I have searched through the issues Issues for a duplicate
• I’ve tested that my Keras model works as a stand-alone

>>> talos.__version__
'0.6.0'

Expected behavior

scan_object.best_model(metric="acc") should result in a new instance of the best performing model.

Actual behavior

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-29-2e96fabca5ec> in <module>
      1 #ta.Deploy(t,"U-nets", metric="acc")
----> 2 scan_object.best_model(metric="acc")

~/.pyenv/versions/miniconda3-4.3.30/envs/tf_gpu/lib/python3.6/site-packages/talos/scan/scan_addon.py in func_best_model(scan_object, metric, asc)
     12     from ..utils.best_model import best_model, activate_model
     13     model_no = best_model(scan_object, metric, asc)
---> 14     out = activate_model(scan_object, model_no)
     15 
     16     return out

~/.pyenv/versions/miniconda3-4.3.30/envs/tf_gpu/lib/python3.6/site-packages/talos/utils/best_model.py in activate_model(self, model_id)
     18     '''Loads the model from the json that is stored in the Scan object'''
     19 
---> 20     model = model_from_json(self.saved_models[model_id])
     21     model.set_weights(self.saved_weights[model_id])
     22 

~/.pyenv/versions/miniconda3-4.3.30/envs/tf_gpu/lib/python3.6/site-packages/keras/engine/saving.py in model_from_json(json_string, custom_objects)
    490     config = json.loads(json_string)
    491     from ..layers import deserialize
--> 492     return deserialize(config, custom_objects=custom_objects)
    493 
    494 

~/.pyenv/versions/miniconda3-4.3.30/envs/tf_gpu/lib/python3.6/site-packages/keras/layers/__init__.py in deserialize(config, custom_objects)
     53                                     module_objects=globs,
     54                                     custom_objects=custom_objects,
---> 55                                     printable_module_name='layer')

~/.pyenv/versions/miniconda3-4.3.30/envs/tf_gpu/lib/python3.6/site-packages/keras/utils/generic_utils.py in deserialize_keras_object(identifier, module_objects, custom_objects, printable_module_name)
    143                     config['config'],
    144                     custom_objects=dict(list(_GLOBAL_CUSTOM_OBJECTS.items()) +
--> 145                                         list(custom_objects.items())))
    146             with CustomObjectScope(custom_objects):
    147                 return cls.from_config(config['config'])

~/.pyenv/versions/miniconda3-4.3.30/envs/tf_gpu/lib/python3.6/site-packages/keras/engine/network.py in from_config(cls, config, custom_objects)
   1020         # First, we create all layers and enqueue nodes to be processed
   1021         for layer_data in config['layers']:
-> 1022             process_layer(layer_data)
   1023         # Then we process nodes in order of layer depth.
   1024         # Nodes that cannot yet be processed (if the inbound node

~/.pyenv/versions/miniconda3-4.3.30/envs/tf_gpu/lib/python3.6/site-packages/keras/engine/network.py in process_layer(layer_data)
   1006 
   1007             layer = deserialize_layer(layer_data,
-> 1008                                       custom_objects=custom_objects)
   1009             created_layers[layer_name] = layer
   1010 

~/.pyenv/versions/miniconda3-4.3.30/envs/tf_gpu/lib/python3.6/site-packages/keras/layers/__init__.py in deserialize(config, custom_objects)
     53                                     module_objects=globs,
     54                                     custom_objects=custom_objects,
---> 55                                     printable_module_name='layer')

~/.pyenv/versions/miniconda3-4.3.30/envs/tf_gpu/lib/python3.6/site-packages/keras/utils/generic_utils.py in deserialize_keras_object(identifier, module_objects, custom_objects, printable_module_name)
    136             if cls is None:
    137                 raise ValueError('Unknown ' + printable_module_name +
--> 138                                  ': ' + class_name)
    139         if hasattr(cls, 'from_config'):
    140             custom_objects = custom_objects or {}

ValueError: Unknown layer: Swish

Model details

MWE

import keras
from keras.models import Model
from keras.layers import Input, Conv2D, BatchNormalization
from keras.layers.advanced_activations import ReLU
import talos as ta

def u_net(shape, nb_filters=64, conv_size=3, init="glorot_uniform",
          activation=ReLU, output_channels=5):

    i = Input(shape, name="input_layer")
    n = Conv2D(nb_filters, conv_size, padding="same", kernel_initializer=init,
        name="block1_conv1")(i)
    n = activation(name="block1_{}1".format(activation.__name__))(n)
    n = BatchNormalization(name="block1_bn1")(n)
    n = Conv2D(nb_filters, conv_size, padding="same", kernel_initializer=init,
        name="block1_conv2")(n)
    n = activation(name="block1_{}2".format(activation.__name__))(n)
    n = BatchNormalization(name="block1_bn2")(n)
    o = Conv2D(output_channels, 1, activation="softmax", name="conv_out")(n)
    return Model(inputs=i, outputs=o)

def talos_model():
    model = u_net(SHAPE, nb_filters=p["nb_filters"], activation=p["act"])
    model.compile(optimizer=p["opt"](lr=1e-4))
    history = model.fit(x=X, y=Y)
    return model, history

scan_object = ta.Scan(x=X, y=Y, model=talos_model, params=p)

parameter dictionary

# fit params
from keras.optimizers import Adam
from keras.layers.advanced_activations import ReLU
from keras_radam import RAdam
from keras_contrib.layers.advanced_activations.swish import Swish

p = {
    "nb_filters": [12, 16, 32],
    "act": [Swish, ReLU],
    "opt": [RAdam, Adam]
}

• My bug report includes an input model
• My bug report includes a parameter dictionary
• My bug report includes a Scan() command
• My bug report question includes a link to a sample of the data

I chose to leave out sample data because it is not relevant to the issue at hand. For the same reason, I chose to create a Minimal Working Example rather than pasting the entire model, which is quite complicated and does not help to locate the issue.

[bjtho08]

Awesome! I’ll be looking forward to seeing the development and maybe pitch in, if I manage to find the time between finishing my phd dissertation and getting my data out of my models 😉

[mikkokotila]

@bjtho08 definitely, that’s a great idea 😃 I’ll change the title of the issue to correspond this scope.