Transparent Keras aims to provide a very simple way to look under the hood during training of Keras models by defining an extra set of outputs that will be returned by train_on_batch or test_on_batch.
The API is extremely simple all that is provided is a TransparentModel that accepts an extra constructor keyword argument observed_tensors. The created model should behave exactly like a Keras model except for the functions (train|test)_on_batch, which return the extra tensors as after their normal return values.
Example
from keras.layers import Activation, Dense, Dropout, Input
import numpy as np
from transparent_keras import TransparentModel
x0 = Input(shape=(10,))
x = Dense(10, activation="relu")(x0)
x = Dropout(0.5)(x)
y_extra = x = Dense(10)(x)
x = Activation("relu")(x)
x = Dropout(0.5)(x)
y = Dense(1)(x)
m = TransparentModel(inputs=[x0], outputs=[y], observed_tensors=[y_extra])
m.compile(optimizer="sgd", loss="mse")
x_random = np.random.rand(128, 10)
y_random = np.random.rand(128, 1)
loss, y_extra = m.train_on_batch(x_random, y_random)1 maintainer
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
File details
Details for the file transparent-keras-0.3.1.tar.gz.
File metadata
- Download URL: transparent-keras-0.3.1.tar.gz
- Upload date:
- Size: 4.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
c5f866c8f74b9ef4b634e5f44ddbb64963536fcc4bd7751a6651a5d90faf83d7
|
|
| MD5 |
131d2a53125b4fd815db77ec2e70e622
|
|
| BLAKE2b-256 |
cbe20a55ab3e97587c039e3996009e75c4f603b22fe380888c4379b9013a4391
|
