Problems with saving custom model for TF serving - tensorflow

I defined simple custom model:
import tensorflow as tf
class CustomModule(tf.keras.layers.Layer):
def __init__(self):
super(CustomModule, self).__init__()
self.v = tf.Variable(1.)
def call(self, x):
print('Tracing with', x)
return x * self.v
def mutate(self, new_v):
self.v.assign(new_v)
I want to save it for serving and that is why I need to provide a function for “serving_default”. I’ve tried to do it like this:
module = CustomModule() module_with_signature_path = './tmp/1' call = tf.function(module.mutate, input_signature=[tf.TensorSpec([], tf.float32)]) tf.saved_model.save(module, module_with_signature_path, signatures=call)
I got an error:
ValueError: Got a non-Tensor value <tf.Operation 'StatefulPartitionedCall' type=StatefulPartitionedCall> for key 'output_0' in the output of the function __inference_mutate_8 used to generate the SavedModel signature 'serving_default'. Outputs for functions used as signatures must be a single Tensor, a sequence of Tensors, or a dictionary from string to Tensor.
How can I properly define signature while saving model? Thank you!

Related

How do I log the outputs of hidden Keras layer during training?

I want to store the float values in the outputs of a particular hidden layer during training. However, since the outputs are KerasTensor objects, I am unable to access them.
How do I access the float values in the tensors so I can store them for later use?
I am currently trying to do this using a custom Callback:
class HidInps(Callback):
def on_train_batch_end(self, batch, logs=None):
layer_out = self.model.get_layer("hidlyr").output
print(layer_out) # KerasTensor(type_spec=TensorSpec(shape=(None, 3), dtype=tf.float32...
print(keras.backend.get_value(layer_out))
However, since the KerasTensor object provides no .numpy() method, eval() or get_value() can not work and I get the appropriate error:
AttributeError: 'KerasTensor' object has no attribute 'numpy'
You need to use the custom training loop in tensorflow to acheive this thing.
Lets say your model instance is refered by the variable my_model. You can create another custom_model from it as follows:
from tensorflow.keras import Model
hidden_layer = self.model.get_layer("hidlyr")
custom_model = Model(inputs=my_model.inputs, outputs=[hidden_layer.output, my_model.output])
with tf.GradientTape() as t:
layer_op, predictions = custom_model(images)
print(layer_op)
For further details, refer https://www.tensorflow.org/tutorials/customization/custom_training_walkthrough#train_the_model
from tensorflow.keras import Model
hidden_layer = self.model.get_layer("hidlyr")
custom_model = Model(inputs=my_model.inputs, outputs=[hidden_layer.output, my_model.output])
with tf.GradientTape() as t:
layer_op, predictions = custom_model(images)
print(layer_op)

How to adopt tensorflow 2 saved model into a module?

All answers I can find related to adopting saved model as part of another model are for keras. but here, I have saved a tf.Module and want to restore it and use it as part of a new model. The saving and loading of the adoptee works, but when saving the adopter, I get an error. Here are the example codes:
from keras.layers import Dense
import tensorflow as tf
import numpy as np
def load_model_safely(path_to_saved_model, signature='serving_default') \
-> tf.types.experimental.ConcreteFunction:
saved_model = tf.saved_model.load(path_to_saved_model)
model = saved_model.signatures[signature]
model._backref_to_saved_model = saved_model
return model
class AdoptedModule(tf.Module):
def __init__(self):
super().__init__()
self.dummy = Dense(3, activation='relu')
#tf.function
def __call__(self, x):
return self.dummy(x)
class AdopterModule(tf.Module):
def __init__(self, adopted):
super().__init__()
self.dummy = adopted
#tf.function
def __call__(self, x):
return self.dummy(x)
adopted = AdoptedModule()
adopted(tf.constant([np.arange(0, 10, dtype=np.float32)]))
tf.saved_model.save(adopted, '/tmp/adopted.tf2',
signatures=adopted.__call__.get_concrete_function(tf.TensorSpec((None, 10), dtype=tf.float32)))
loaded_adopted = load_model_safely('/tmp/adopted.tf2')
adopter = AdopterModule(adopted=loaded_adopted)
adopter(tf.constant([np.arange(0, 10, dtype=np.float32)]))
tf.saved_model.save(adopter, '/tmp/adopter.tf2',
signatures=adopted.__call__.get_concrete_function(tf.TensorSpec((None, 10), dtype=tf.float32)))
The error is
AssertionError: Tried to export a function which references 'untracked' resource Tensor("124629:0", shape=(), dtype=resource). TensorFlow objects (e.g. tf.Variable) captured by functions must be 'tracked' by assigning them to an attribute of a tracked object or assigned to an attribute of the main object directly. See the information below:
Function name = b'__inference_signature_wrapper_124635'
Captured Tensor = <ResourceHandle(name="Resource-289-at-0x600028cdc500", device="/job:localhost/replica:0/task:0/device:CPU:0", container="Anonymous", type="tensorflow::Var", dtype and shapes : "[ DType enum: 1, Shape: [10,3] ]")>
Trackable referencing this tensor = <tf.Variable 'dense_3/kernel:0' shape=(10, 3) dtype=float32>
I can probably hack it by first recreating the adopted module instance (by calling the original codes with exact hyperparameters) and then force-assign all trainable variables from the loaded concrete function. But one may not even have the original model codes and just want to adopt the trained autograph. Further, the adopted layer should be trainable during fine-tuning the whole model.

how to use restored model with tensorflow 2.0

The model is built as
class MyModel(tf.keras.Model):
def __init__(self, num_states, hidden_units, num_actions):
super(MyModel, self).__init__()
## btach_size * size_state
self.input_layer = tf.keras.layers.InputLayer(input_shape=(num_states,))
self.hidden_layers = []
for i in hidden_units:
self.hidden_layers.append(tf.keras.layers.Dense(
i, activation='tanh', kernel_initializer='RandomNormal'))
self.output_layer = tf.keras.layers.Dense(
num_actions, activation='linear', kernel_initializer='RandomNormal')
#tf.function
def call(self, inputs):
z = self.input_layer(inputs)
for layer in self.hidden_layers:
z = layer(z)
output = self.output_layer(z)
return output
after training, the model was saved as:
tf.saved_model.save(self.model, model_path_dir='saved_model/model_checkpoint')
Now, I'm trying to restore the model and then do inference as:
train_agent = tf.saved_model.load('saved_model/model_checkpoint')
## input_state is an array with shape of (num_states, )
action = train_agent(np.atleast_2d(input_state))
However, I keep getting this error:
TypeError: '_UserObject' object is not callable
How can I exactly use the restored model?
I haven't figure out how to use the tf.saved_model.save() and tf.saved_model.load() properly, if anyone could help, it would be great! Current solution is to use save_weights() instead. For example:
model = Model()
mode.train()
...
model.save_weights(checkpoint_dir)
input_s = np.random.normal(1, 128)
output = model(input_s) ## output a tensor
You correctly decorated the call method with tf.function, thus, your SavedModel will have the very same method serialized inside.
Thus, you can just call your model by calling the call method:
action = train_agent.call(np.atleast_2d(input_state))
Required code given below
tf.keras.models.save_model(model_name,path)
import tensorflow as tf
#save the model for later use
tf.keras.models.save_model(classifier_model,'F:/Face_Recognition/face_classifier_model.h5')
#Reload the model
classifier_model=tf.keras.models.load_model('F:/Face_Recognition/face_classifier_model.h5')

Simple Custom Layer fails to build in simple model

When using a custom class to replace a lambda function, building a model fails.
I previously had this code in a lambda function and it worked fine but I was unable to save a model. I need to save the model that I'm building which has dependencies on this code snippet.
import keras
import tensorflow as tf
class ShapePositionLayer(keras.layers.Layer):
def call(self, x):
assert isinstance(x, list)
a, b = x
return keras.backend.gather(keras.backend.shape(a), b)
def compute_output_shape(self, input_shape):
return (1)
captions = keras.layers.Input(shape=[5,1024], name='captions')
batch_size = ShapePositionLayer()([captions,tf.constant(0,
dtype=tf.int32)])
model = keras.models.Model(inputs=[captions], outputs=[batch_size])
I expected to be able to build a model.
Instead receive error:
AttributeError: 'NoneType' object has no attribute '_inbound_nodes'

How to build this custom layer in Keras?

I'm building a NN that supports complex numbers. Currently working on complex activation. According to a Benjio paper, this is a good one:
Where b is a trainable parameter to be learnt. So I'm building a special layer to do this activation. I'm new to Keras and stuck already. I created this code below, but it gives an error with the build function. I have no idea what's happening, I just tried to copy the template. Please help.
class modrelu(Layer):
def __init__(self, **kwargs):
super(modrelu, self).__init__(**kwargs)
def build(self):
self.b= K.variable(value=np.random.rand()-0.5, dtype='float64')
super(modrelu, self).build() # Be sure to call this at the end
def call(self, x):
assert isinstance(x, list)
ip_r, ip_i = x
comp= tf.complex(ip_r, ip_i )
ABS= tf.math.abs(comp)
ANG= tf.math.angle(comp)
ABS= K.relu( self.b + ABS)
op_r= ABS * K.sin(angle) #K.dot ??
op_i= ABS * K.cos(angle)
return [op_r, op_i]
def compute_output_shape(self, input_shape):
assert isinstance(input_shape, list)
shape_a, shape_b = input_shape
return [shape_a, shape_b]
Comments on my code:
In the init I didn't add anything, cause it is an activation layer that takes no input when instantiated.
In the build method, I tried to add the b's. Not sure if I should use the self.add_weight method. Ideally, I want to have as many b's as the dimension of input.
In the call method, this one, I'm pretty sure what I'm doing. It is easy, I just implemented the function.
The last one, compute_output_shape, I just copied-pasted the template. The output should be the same as the input, cause it is just an activation layer.
Finally, the error for what its worth, I know it is nonsense
TypeError Traceback (most recent call last)
<ipython-input-5-3101a9226da5> in <module>
1 a=K.variable(np.array([1,2]))
2 b=K.variable(np.array([3,4]))
----> 3 act([a,b])
~\AppData\Local\conda\conda\envs\python36\lib\site-packages\keras\engine\base_layer.py in __call__(self, inputs, **kwargs)
429 'You can build it manually via: '
430 '`layer.build(batch_input_shape)`')
--> 431 self.build(unpack_singleton(input_shapes))
432 self.built = True
433
TypeError: build() takes 1 positional argument but 2 were given
There are several issues with your code.
First of all I should address the error you get from interpreter:
TypeError: build() takes 1 positional argument but 2 were given
The build method should take input_shape argument. Therefore you should declare build method as build(self, input_shape)
The second issue is undefined shape of the variables in the build method. You should explicitly declare shape of the variables. In your case the np.random.rand array should be of input_shape shape.
Another issue is that you are trying to return 2 results ([op_r, op_i]) in the call method. I'm not specialist in Keras but as far as I know you can't do it. Every Keras layer should have one and only one output. See here for the details: https://github.com/keras-team/keras/issues/3061
However if you use tensorflow backend you may use complex numbers (tf.complex) to return both real (op_r) and imagenary (op_i) parts of the complex number.
Here is the working implementation of modrelu layer with simple usage example. It is writtern for TensorFlow 1.12.0 which is distributed with it's own implementation of Keras API but I think you can easily adopt it for original Keras:
import tensorflow as tf
from tensorflow.python.keras import backend as K
from tensorflow.python.keras.engine import Layer
import numpy as np
class modrelu(Layer):
def __init__(self, **kwargs):
super(modrelu, self).__init__(**kwargs)
# provide input_shape argument in the build method
def build(self, input_shape):
# You should pass shape for your variable
self.b= K.variable(value=np.random.rand(*input_shape)-0.5,
dtype='float32')
super(modrelu, self).build(input_shape) # Be sure to call this at the end
def call(self, inputs, **kwargs):
assert inputs.dtype == tf.complex64
ip_r = tf.math.real(inputs)
ip_i = tf.math.imag(inputs)
comp = tf.complex(ip_r, ip_i )
ABS = tf.math.abs(comp)
ANG = tf.math.angle(comp)
ABS = K.relu(self.b + ABS)
op_r = ABS * K.sin(ANG) #K.dot ??
op_i = ABS * K.cos(ANG)
# return single tensor in the call method
return tf.complex(op_r, op_i)
real = tf.constant([2.25, 3.25])
imag = tf.constant([4.75, 5.75])
x = tf.complex(real, imag)
y = modrelu()(x)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
print(sess.run(y))
P.S.: I didn't check the math so you should check it by yourself.
You are not coding the layer correctly, the build function takes a input_shape parameter, which you can use to initialize the weights/parameters of your layer.
You can see an example in Keras' source code.