How do I add the None Dimension back to a Tensor?
.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty{ height:90px;width:728px;box-sizing:border-box;
}
I've made some transformation in a Lambda Layer and now I have shape (1,), how do I get back to (None, 1)?
Here are my operations
def function_lambda(x):
import keras.backend
aux_array = keras.backend.sign(x)
#the shape is (?, 11) here OK
aux_array = keras.backend.relu(aux_array)
#the shape is (?, 11) here still OK
aux_array = keras.backend.any(aux_array)
#the shape is () here not OK
aux_array = keras.backend.reshape(aux_array, [-1])
#now the shape is (1,) almost OK
return aux_array
How do I reshape and put back the None Batch Dimension?
python tensorflow keras reshape
asked Nov 23 '18 at 8:17
ViniciusVinicius
158
add a comment |
I've made some transformation in a Lambda Layer and now I have shape (1,), how do I get back to (None, 1)?
Here are my operations
def function_lambda(x):
import keras.backend
aux_array = keras.backend.sign(x)
#the shape is (?, 11) here OK
aux_array = keras.backend.relu(aux_array)
#the shape is (?, 11) here still OK
aux_array = keras.backend.any(aux_array)
#the shape is () here not OK
aux_array = keras.backend.reshape(aux_array, [-1])
#now the shape is (1,) almost OK
return aux_array
How do I reshape and put back the None Batch Dimension?
python tensorflow keras reshape
asked Nov 23 '18 at 8:17
ViniciusVinicius
158
did you figure it out?
– enjal
Apr 1 at 21:43
add a comment |
I've made some transformation in a Lambda Layer and now I have shape (1,), how do I get back to (None, 1)?
Here are my operations
def function_lambda(x):
import keras.backend
aux_array = keras.backend.sign(x)
#the shape is (?, 11) here OK
aux_array = keras.backend.relu(aux_array)
#the shape is (?, 11) here still OK
aux_array = keras.backend.any(aux_array)
#the shape is () here not OK
aux_array = keras.backend.reshape(aux_array, [-1])
#now the shape is (1,) almost OK
return aux_array
How do I reshape and put back the None Batch Dimension?
python tensorflow keras reshape
asked Nov 23 '18 at 8:17
ViniciusVinicius
158
I've made some transformation in a Lambda Layer and now I have shape (1,), how do I get back to (None, 1)?
Here are my operations
def function_lambda(x):
import keras.backend
aux_array = keras.backend.sign(x)
#the shape is (?, 11) here OK
aux_array = keras.backend.relu(aux_array)
#the shape is (?, 11) here still OK
aux_array = keras.backend.any(aux_array)
#the shape is () here not OK
aux_array = keras.backend.reshape(aux_array, [-1])
#now the shape is (1,) almost OK
return aux_array
How do I reshape and put back the None Batch Dimension?
python tensorflow keras reshape
python tensorflow keras reshape
asked Nov 23 '18 at 8:17
ViniciusVinicius
158
asked Nov 23 '18 at 8:17
ViniciusVinicius
158
asked Nov 23 '18 at 8:17
ViniciusVinicius
158
asked Nov 23 '18 at 8:17
ViniciusVinicius
158
asked Nov 23 '18 at 8:17
ViniciusVinicius
158
158
did you figure it out?
– enjal
Apr 1 at 21:43
add a comment |
did you figure it out?
– enjal
Apr 1 at 21:43
did you figure it out?
– enjal
Apr 1 at 21:43
did you figure it out?
– enjal
Apr 1 at 21:43
add a comment |
2 Answers
2
active
oldest
votes
If you have a fully-defined shape (like (1,)) then you don't need None dimensions, because you know exactly how many elements your tensor has. But you can reshape to have two dimensions replacing you reshaping with:
aux_array = keras.backend.reshape(aux_array, [-1, 1])
This will give you an array with shape (1, 1), which is compatible with the shape (None, 1), so it should be okay. Note that -1 in reshape means "however many elements are necessary to accommodate the tensor in this shape", but in this case you know you only have one element, so it would be the same to use:
aux_array = keras.backend.reshape(aux_array, [1, 1])
Because, again, the shape is fully defined and you know what the size of each dimension should be exactly. However, using -1 is convenient because it works whether you know the shape fully or not, and Keras/TensorFlow will work out what the size should be (the dimension will have the necessary size if it can be computed or None if part of the shape is unknown).
answered Nov 23 '18 at 12:08
jdehesajdehesa
27.8k43759
add a comment |
The tf.reshape has a limitation that it cannot work with "None" dimension. So far as I know, the only place where we can define a "None" dimension is in tf.placeholder. The following should work:
def function_lambda(x):
import keras.backend
aux_array = keras.backend.sign(x)
#the shape is (?, 11) here OK
aux_array = keras.backend.relu(aux_array)
#the shape is (?, 11) here still OK
aux_array = keras.backend.any(aux_array)
#the shape is () here not OK
aux_array = keras.backend.placeholder_with_default(aux_array, [None,1])
#now the shape is (?,1) OK
return aux_array
Additional note: creating a new placeholder for reintroducing the "None" dimension is useful when trying to use Google Machine Learning Engine. MLE requires that the first dimension (batch) always remain "None" or unknown.
answered Mar 31 at 3:01
rahullakrahullak
12
add a comment |
Your Answer
StackExchange.ifUsing("editor", function () {
StackExchange.using("externalEditor", function () {
StackExchange.using("snippets", function () {
StackExchange.snippets.init();
});
});
}, "code-snippets");
StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "1"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);
StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});
function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});
}
});
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53442887%2fhow-do-i-add-the-none-dimension-back-to-a-tensor%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
2 Answers
2
active
oldest
votes
2 Answers
2
active
oldest
votes
active
oldest
votes
active
oldest
votes
If you have a fully-defined shape (like (1,)) then you don't need None dimensions, because you know exactly how many elements your tensor has. But you can reshape to have two dimensions replacing you reshaping with:
aux_array = keras.backend.reshape(aux_array, [-1, 1])
This will give you an array with shape (1, 1), which is compatible with the shape (None, 1), so it should be okay. Note that -1 in reshape means "however many elements are necessary to accommodate the tensor in this shape", but in this case you know you only have one element, so it would be the same to use:
aux_array = keras.backend.reshape(aux_array, [1, 1])
Because, again, the shape is fully defined and you know what the size of each dimension should be exactly. However, using -1 is convenient because it works whether you know the shape fully or not, and Keras/TensorFlow will work out what the size should be (the dimension will have the necessary size if it can be computed or None if part of the shape is unknown).
answered Nov 23 '18 at 12:08
jdehesajdehesa
27.8k43759
add a comment |
If you have a fully-defined shape (like (1,)) then you don't need None dimensions, because you know exactly how many elements your tensor has. But you can reshape to have two dimensions replacing you reshaping with:
aux_array = keras.backend.reshape(aux_array, [-1, 1])
This will give you an array with shape (1, 1), which is compatible with the shape (None, 1), so it should be okay. Note that -1 in reshape means "however many elements are necessary to accommodate the tensor in this shape", but in this case you know you only have one element, so it would be the same to use:
aux_array = keras.backend.reshape(aux_array, [1, 1])
Because, again, the shape is fully defined and you know what the size of each dimension should be exactly. However, using -1 is convenient because it works whether you know the shape fully or not, and Keras/TensorFlow will work out what the size should be (the dimension will have the necessary size if it can be computed or None if part of the shape is unknown).
answered Nov 23 '18 at 12:08
jdehesajdehesa
27.8k43759
add a comment |
If you have a fully-defined shape (like (1,)) then you don't need None dimensions, because you know exactly how many elements your tensor has. But you can reshape to have two dimensions replacing you reshaping with:
aux_array = keras.backend.reshape(aux_array, [-1, 1])
This will give you an array with shape (1, 1), which is compatible with the shape (None, 1), so it should be okay. Note that -1 in reshape means "however many elements are necessary to accommodate the tensor in this shape", but in this case you know you only have one element, so it would be the same to use:
aux_array = keras.backend.reshape(aux_array, [1, 1])
Because, again, the shape is fully defined and you know what the size of each dimension should be exactly. However, using -1 is convenient because it works whether you know the shape fully or not, and Keras/TensorFlow will work out what the size should be (the dimension will have the necessary size if it can be computed or None if part of the shape is unknown).
answered Nov 23 '18 at 12:08
jdehesajdehesa
27.8k43759
If you have a fully-defined shape (like (1,)) then you don't need None dimensions, because you know exactly how many elements your tensor has. But you can reshape to have two dimensions replacing you reshaping with:
aux_array = keras.backend.reshape(aux_array, [-1, 1])
This will give you an array with shape (1, 1), which is compatible with the shape (None, 1), so it should be okay. Note that -1 in reshape means "however many elements are necessary to accommodate the tensor in this shape", but in this case you know you only have one element, so it would be the same to use:
aux_array = keras.backend.reshape(aux_array, [1, 1])
Because, again, the shape is fully defined and you know what the size of each dimension should be exactly. However, using -1 is convenient because it works whether you know the shape fully or not, and Keras/TensorFlow will work out what the size should be (the dimension will have the necessary size if it can be computed or None if part of the shape is unknown).
answered Nov 23 '18 at 12:08
jdehesajdehesa
27.8k43759
answered Nov 23 '18 at 12:08
jdehesajdehesa
27.8k43759
answered Nov 23 '18 at 12:08
jdehesajdehesa
27.8k43759
answered Nov 23 '18 at 12:08
jdehesajdehesa
27.8k43759
27.8k43759
add a comment |
add a comment |
The tf.reshape has a limitation that it cannot work with "None" dimension. So far as I know, the only place where we can define a "None" dimension is in tf.placeholder. The following should work:
def function_lambda(x):
import keras.backend
aux_array = keras.backend.sign(x)
#the shape is (?, 11) here OK
aux_array = keras.backend.relu(aux_array)
#the shape is (?, 11) here still OK
aux_array = keras.backend.any(aux_array)
#the shape is () here not OK
aux_array = keras.backend.placeholder_with_default(aux_array, [None,1])
#now the shape is (?,1) OK
return aux_array
Additional note: creating a new placeholder for reintroducing the "None" dimension is useful when trying to use Google Machine Learning Engine. MLE requires that the first dimension (batch) always remain "None" or unknown.
answered Mar 31 at 3:01
rahullakrahullak
12
add a comment |
The tf.reshape has a limitation that it cannot work with "None" dimension. So far as I know, the only place where we can define a "None" dimension is in tf.placeholder. The following should work:
def function_lambda(x):
import keras.backend
aux_array = keras.backend.sign(x)
#the shape is (?, 11) here OK
aux_array = keras.backend.relu(aux_array)
#the shape is (?, 11) here still OK
aux_array = keras.backend.any(aux_array)
#the shape is () here not OK
aux_array = keras.backend.placeholder_with_default(aux_array, [None,1])
#now the shape is (?,1) OK
return aux_array
Additional note: creating a new placeholder for reintroducing the "None" dimension is useful when trying to use Google Machine Learning Engine. MLE requires that the first dimension (batch) always remain "None" or unknown.
answered Mar 31 at 3:01
rahullakrahullak
12
add a comment |
The tf.reshape has a limitation that it cannot work with "None" dimension. So far as I know, the only place where we can define a "None" dimension is in tf.placeholder. The following should work:
def function_lambda(x):
import keras.backend
aux_array = keras.backend.sign(x)
#the shape is (?, 11) here OK
aux_array = keras.backend.relu(aux_array)
#the shape is (?, 11) here still OK
aux_array = keras.backend.any(aux_array)
#the shape is () here not OK
aux_array = keras.backend.placeholder_with_default(aux_array, [None,1])
#now the shape is (?,1) OK
return aux_array
Additional note: creating a new placeholder for reintroducing the "None" dimension is useful when trying to use Google Machine Learning Engine. MLE requires that the first dimension (batch) always remain "None" or unknown.
answered Mar 31 at 3:01
rahullakrahullak
12
The tf.reshape has a limitation that it cannot work with "None" dimension. So far as I know, the only place where we can define a "None" dimension is in tf.placeholder. The following should work:
def function_lambda(x):
import keras.backend
aux_array = keras.backend.sign(x)
#the shape is (?, 11) here OK
aux_array = keras.backend.relu(aux_array)
#the shape is (?, 11) here still OK
aux_array = keras.backend.any(aux_array)
#the shape is () here not OK
aux_array = keras.backend.placeholder_with_default(aux_array, [None,1])
#now the shape is (?,1) OK
return aux_array
Additional note: creating a new placeholder for reintroducing the "None" dimension is useful when trying to use Google Machine Learning Engine. MLE requires that the first dimension (batch) always remain "None" or unknown.
answered Mar 31 at 3:01
rahullakrahullak
12
answered Mar 31 at 3:01
rahullakrahullak
12
answered Mar 31 at 3:01
rahullakrahullak
12
answered Mar 31 at 3:01
rahullakrahullak
12
12
add a comment |
add a comment |
Thanks for contributing an answer to Stack Overflow!
- Please be sure to answer the question. Provide details and share your research!
But avoid …
- Asking for help, clarification, or responding to other answers.
- Making statements based on opinion; back them up with references or personal experience.
To learn more, see our tips on writing great answers.
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53442887%2fhow-do-i-add-the-none-dimension-back-to-a-tensor%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
did you figure it out?
– enjal
Apr 1 at 21:43