use cntk trained model with python - python

I have trained a model using CNTK, lets call simple.dnn
now for the phase of testing I do not want to install CNTK on windows,but use trained model with python. How can I use trained model (weights,...) for testing using python?

You can use the load_model function, see https://www.cntk.ai/pythondocs/cntk.html?highlight=load_model#cntk.persist.load_model. The basic flow should look like this:
from cntk import load_model
loaded_model = load_model("yourModel.model", 'float')
output = model.eval(arguments)

Related

import model from ML.NET to Python Keras

I want to load a model trained in ML.NET in Python Keras to try and see the result. The model file is zip formatted and I don't have train data. How can I do that?

Error while importing VGG16 h5 file ValueError: No model found in config file

I tried to import vgg16 which I downloaded from google storage
import keras
import cv2
from keras.models import Sequential, load_model
But I got that error
ValueError: No model found in config file.
I was able to recreate the issue using your code and downloaded weights file mentioned by you. I am not sure about the reason for the issue but I can offer an alternative way for you to use pretrained vgg16 model from keras.
You need to use model from keras.applications file
Here is the link for your reference https://keras.io/api/applications/
There are three ways to instantiate this model by using weights argument which takes any of following three values None/'imagenet'/filepathToWeightsFile.
Since you have already downloaded the weights , I suggest that you use the filepath option like the below code but for first time usage I will suggest to use imagenet (option 3). It will download the weight file which can be saved and reused later.
You need to add the following lines of code.
Option 1:
from keras.applications.vgg16 import VGG16
model = VGG16(weights = 'vgg16_weights_tf_dim_ordering_tf_kernels.h5')
Option 2:
from keras.applications.vgg16 import VGG16
model = VGG16(weights = None)
model.load_weights('vgg16_weights_tf_dim_ordering_tf_kernels.h5')
Option 3: for using pretrained imagenet weights
from keras.applications.vgg16 import VGG16
model = VGG16(weights = 'imagenet')
The constructor also takes other arguments like include_top etc which can be added as per requirement.
The problem here is that you're trying to load a model that is not a model and probably are just weights: so the problem is not in the load of the model but in the save.
When you are saving the model try:
If you are using callbacks then "save_weights_only"=False
Else use the function tf.keras.models.save_model(model,filepath)
A complete model has two parts: model architecture and weights.
So if we just have weights, we must first load architecture(may be use python file or keras file ), and then load weights on the architecture.
for example:
model = tf.keras.models.load_model("facenet_keras.h5")
model.load_weights("facenet_keras_weights.h5")

tf.keras manual device placement

Migrating to the TF2.0 I'm trying to use the tf.keras approach for solving things.
In standard TF, I can use with tf.device(...) to control where ops are.
For example, I might have a model something like
model = tf.keras.Sequential([tf.keras.layers.Input(..),
tf.keras.layers.Embedding(...),
tf.keras.layers.LSTM(...),
...])
Assuming I want to have the network up until Embedding (including) on the CPU and the and from there on on the GPU, how will I go about that?
(This is just an example, the layers could have nothing to do with embeddings)
If the solution involves subclassing tf.keras.Model that is OK too, I don't mind not using Sequential
You can use the Keras functional API:
inputs = tf.keras.layers.Input(..)
with tf.device("/GPU:0"):
model = tf.keras.layers.Embedding(...)(inputs)
outputs = tf.keras.layers.LSTM(...)(model)
model = tf.keras.Model(inputs=inputs, outputs=outputs)

How to distribute trained model?

I want to distribute my trained deep-learning model which is made using Keras.
The partner want to use it without complicated environment construction, such as installing python, keras, etc...
How can I do this? Should I make an .exe file?
Maybe you want to look into this tutoiral.
Simplest code possible:
from keras.models import load_model
model.save('my_model.h5') # creates a HDF5 file 'my_model.h5'
# returns a compiled model
# identical to the previous one
model = load_model('my_model.h5')

Using Keras serialized model with dropout in pyspark

I have several neural networks built using Keras that I used so far mostly in Jupyter. I often save models from scikit-learn with joblib and Keras with json + hdf5 and use them in other notebooks without issue.
I made a Python Spark application that can make use of those serialized models in cluster mode. joblib models are working fine however, I encountered an issue with Keras.
Here is the model used in notebook and pyspark:
def build_gru_model():
model = Sequential()
model.add(Embedding(max_nb_words, 128, input_length=max_sequence_length, dropout=0.2))
model.add(GRU(128, dropout_W=0.2, dropout_U=0.2))
model.add(Dense(2, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
return model
both called the same way:
preds = model.predict_proba(data, verbose=0)
However, only in Spark I get the error:
MissingInputError: ("An input of the graph, used to compute DimShuffle{x,x,x,x}(keras_learning_phase), was not provided and not given a value.Use the Theano flag exception_verbosity='high',for more information on this error.", keras_learning_phase)
I've done the mandatory search and found: https://github.com/fchollet/keras/issues/2430 which points to https://keras.io/getting-started/faq/
If I indeed remove dropout from my model, it works. However, I fail to understand how to implement something that would allow me to keep dropout during the training phase like described in the FAQ.
Based on the model code, how one would accomplish this?
You can try to put (before your prediction)
import keras.backend as K
K.set_learning_phase(0)
It should set your learning phase to 0 (test time)

Categories

Resources