experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Corrected order of parameters in keras WideDeepModel

Browse Source
This commit is contained in:
Ashutosh Hathidara 2020-03-12 19:25:06 +05:30
parent 453186c1a0
commit da14779469
1 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
tensorflow/python/keras/premade/wide_deep.py
View File

@ -42,14 +42,14 @@ class WideDeepModel(keras_training.Model):
linear_model = LinearModel()
dnn_model = keras.Sequential([keras.layers.Dense(units=64),
keras.layers.Dense(units=1)])
combined_model = WideDeepModel(dnn_model, linear_model)
combined_model = WideDeepModel(linear_model, dnn_model)
combined_model.compile(optimizer=['sgd', 'adam'], 'mse', ['mse'])
# define dnn_inputs and linear_inputs as separate numpy arrays or
# a single numpy array if dnn_inputs is same as linear_inputs.
combined_model.fit([dnn_inputs, linear_inputs], y, epochs)
combined_model.fit([linear_inputs, dnn_inputs], y, epochs)
# or define a single `tf.data.Dataset` that contains a single tensor or
# separate tensors for dnn_inputs and linear_inputs.
dataset = tf.data.Dataset.from_tensors(([dnn_inputs, linear_inputs], y))
dataset = tf.data.Dataset.from_tensors(([linear_inputs, dnn_inputs], y))
combined_model.fit(dataset, epochs)
```
@ -64,9 +64,9 @@ class WideDeepModel(keras_training.Model):
dnn_model = keras.Sequential([keras.layers.Dense(units=1)])
dnn_model.compile('rmsprop', 'mse')
dnn_model.fit(dnn_inputs, y, epochs)
combined_model = WideDeepModel(dnn_model, linear_model)
combined_model = WideDeepModel(linear_model, dnn_model)
combined_model.compile(optimizer=['sgd', 'adam'], 'mse', ['mse'])
combined_model.fit([dnn_inputs, linear_inputs], y, epochs)
combined_model.fit([linear_inputs, dnn_inputs], y, epochs)
```
"""