Saving & Loading CRF layer with load_model

[codertimo]

Now I'm using the CRF layer very usefully. Thanx for developed it :) But I found some inconvenience part, which is saving and loading the model.

I wanted to load the CRF layer with load_model(model_path).

Other layers in keras-contrib can automatically load the model with load_model(). because get_custom_objects().update({'CosineDense': CosineDense}) infrom to origin-keras about the custom layer.

However, the CRF Layer cannot contain this code, because CRF contains custom loss function Like this code

model.compile(loss=crf.loss_function,
               optimizer='adam',
               metrics=[crf.accuracy])

I also tried this code too but it's not working

model.save(path, custom_objects={'CRF':CRF})
load_model(path, custom_objects={'CRF':CRF})

File "model.py", line 16, in __init__
    self.model = load_model(model_path, custom_objects={'CRF': CRF})
  File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/keras/models.py", line 264, in load_model
    sample_weight_mode=sample_weight_mode)
  File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/keras/models.py", line 781, in compile
    **kwargs)
  File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/keras/engine/training.py", line 681, in compile
    loss_function = losses.get(loss)
  File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/keras/losses.py", line 102, in get
    return deserialize(identifier)
  File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/keras/losses.py", line 94, in deserialize
    printable_module_name='loss function')
  File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/keras/utils/generic_utils.py", line 159, in deserialize_keras_object
    ':' + function_name)
ValueError: Unknown loss function:loss

Please give me some idea for saving CRF layer and loss function :)

[codertimo]

The way that I found. But there is some issue.

get_custom_objects().update({'CRF':CRF, 'CRFLoss':CRF.loss_function})

Under code shows the CRF layers loss_function()

@property
    def loss_function(self):
        if self.learn_mode == 'join':
            def loss(y_true, y_pred):
                assert self.inbound_nodes, 'CRF has not connected to any layer.'
                assert not self.outbound_nodes, 'When learn_model="join", CRF must be the last layer.'
                if self.sparse_target:
                    y_true = K.one_hot(K.cast(y_true[:, :, 0], 'int32'), self.units)
                X = self.inbound_nodes[0].input_tensors[0]
                mask = self.inbound_nodes[0].input_masks[0]
                nloglik = self.get_negative_log_likelihood(y_true, X, mask)
                return nloglik

            return loss
        else:
            if self.sparse_target:
                return sparse_categorical_crossentropy
            else:
                return categorical_crossentropy

So loss function is not static (object public function), and it calculates the loss using the own object. Is there any method that can make independent crf_loss_function?

[iliaschalkidis]

@codertimo I think a good solution is given here (https://github.com/farizrahman4u/keras-contrib/issues/125).

[doggeral]

@codertimo have you solved the issue? Can you help to share some hints with me?

[kamei86i]

@codertimo Did you find a viable solution different from #125 ?

[kamei86i]

Taking inspiration from here: https://github.com/Hironsan/keras-crf-layer I think I've found a work-around solution.

to save a model, just use the save method of model... model.save(path)

to reload the model I use this:

from keras_contrib.layers import CRF
from keras.models import load_model

def create_custom_objects():
    instanceHolder = {"instance": None}
    class ClassWrapper(CRF):
        def __init__(self, *args, **kwargs):
            instanceHolder["instance"] = self
            super(ClassWrapper, self).__init__(*args, **kwargs)
    def loss(*args):
        method = getattr(instanceHolder["instance"], "loss_function")
        return method(*args)
    def accuracy(*args):
        method = getattr(instanceHolder["instance"], "accuracy")
        return method(*args)
    return {"ClassWrapper": ClassWrapper ,"CRF": ClassWrapper, "loss": loss, "accuracy":accuracy}

def load_keras_model(path):
    model = load_model(path, custom_objects=create_custom_objects())
    return model

[lzixuan]

@kamei86i Thank you for this solution, and using it to reload the model works well on the "predict" function. However, when I tried to use the "evaluate" after loading the model, it raised an "Incompatible shapes" exception.

[dorothee]

@kamei86i Thanks for the solution. However, when I am loading it using this, the weights are actually randomly initialized. The model evaluation gives random results, each time I load the same file. Can you post the full working solution?

[maryam1990]

@kamei86i Thank you for the solution. i am able to reload the model. However when I tried to continue training the model, it raised an "Incompatible shapes" exception.

Could you help me