Trjtdtk

Why did the EU agree to delay the Brexit deadline?

Can someone explain how this makes sense electrically?

Travelling outside the UK without a passport

Is a model fitted to data or is data fitted to a model?

What if a revenant (monster) gains fire resistance?

Why is it that I can sometimes guess the next note?

Argument list too long when zipping large list of certain files in a folder

GraphicsGrid with a Label for each Column and Row

What should you do if you miss a job interview (deliberately)?

Where did Heinlein say "Once you get to Earth orbit, you're halfway to anywhere in the Solar System"?

Does an advisor owe his/her student anything? Will an advisor keep a PhD student only out of pity?

Melting point of aspirin, contradicting sources

Is there a single word describing earning money through any means?

Can I sign legal documents with a smiley face?

React - map array to child component

How much character growth crosses the line into breaking the character

The Staircase of Paint

How is flyblackbird.com operating under Part 91K?

what is different between Do you interest vs interested in something?

How should I respond when I lied about my education and the company finds out through background check?

Pre-mixing cryogenic fuels and using only one fuel tank

Offered money to buy a house, seller is asking for more to cover gap between their listing and mortgage owed

Multiplicative persistence

When a Cleric spontaneously casts a Cure Light Wounds spell, will a Pearl of Power recover the original spell or Cure Light Wounds?

What is the meaning of ValueError in Keras? - 'Tensor conversion requested dtype complex64 for Tensor with dtype float32'

Keras — Transfer learning — changing Input tensor shape“concat” mode can only merge layers with matching output shapes except for the concat axisKeras shape error in applications Inception Resnet v2How to create a dummy model in TensorflowHow to set input for proper fit with lstm?model.predict in Keras, Python errorUsing Keras masking layer with 2D convolutions (Conv2D)Cannot interpret feed_dict key as Tensor: Tensor Tensor(“Placeholder:0”, shape=(3, 3, 3, 32), dtype=float32) is not an element of this graphKeras/TF: Making sure image training data shape is accurate for Time Distributed CNN+LSTMHow to choose the number of output channels in a convolutional layer?

1

$begingroup$

I have the following part of code -

...
model.add(Generic_act_func())
print('layer 1 output')
print(model.layers[1].output) #Here layers[1] is Generic_act_func() layer specified above
model.add(Lambda(lambda x: tf.spectral.irfft2d(x)))
...

Which when run produces the following output and error respectively -

Output -

layer 1 output
Tensor("generic_act_func_1/Complex:0", shape=(?, 28, 28, 9), dtype=complex64)

Error -

ValueError: Tensor conversion requested dtype complex64 for Tensor with dtype float32: 'Tensor("lambda_1/Placeholder:0", shape=(?, 28, 28, 9), dtype=float32)'
at line

model.add(Lambda(lambda x: tf.spectral.irfft2d(x)))

tf.spectral.irfft2d() requires a complex64 input which the previous layer i.e., Generic_act_func() layer is already producing (as printed by layer 1 output), hence I am not able to figure out why this error. Thanks in advance for any help for solving the error.

Edit - Could there be a mistake wrt passing previous layer outputs to the Lambda layer? Like I have used a variable 'x' in the Lambda function assuming the previous layer output will be captured in this variable. Is this the correct way of adding the Lambda layer? (I am a beginner in Keras, so will be really grateful for any help)

deep-learning keras computer-vision

share|improve this question

edited Mar 20 at 6:35

psj

asked Mar 19 at 13:50

psjpsj

62

New contributor

psj is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  Making a separate custom layer rather than using a Lambda layer solved the issue for me. The possible reason according to me might be the use of 'tf.spectral.irfft' function inside the Lambda function. This function changes the output shape which needs to be specified using compute_output_shape.
  $endgroup$
  – psj
  2 days ago
  

  
  
  
  

add a comment | 

1

$begingroup$

I have the following part of code -

...
model.add(Generic_act_func())
print('layer 1 output')
print(model.layers[1].output) #Here layers[1] is Generic_act_func() layer specified above
model.add(Lambda(lambda x: tf.spectral.irfft2d(x)))
...

Which when run produces the following output and error respectively -

Output -

layer 1 output
Tensor("generic_act_func_1/Complex:0", shape=(?, 28, 28, 9), dtype=complex64)

Error -

ValueError: Tensor conversion requested dtype complex64 for Tensor with dtype float32: 'Tensor("lambda_1/Placeholder:0", shape=(?, 28, 28, 9), dtype=float32)'
at line

model.add(Lambda(lambda x: tf.spectral.irfft2d(x)))

tf.spectral.irfft2d() requires a complex64 input which the previous layer i.e., Generic_act_func() layer is already producing (as printed by layer 1 output), hence I am not able to figure out why this error. Thanks in advance for any help for solving the error.

Edit - Could there be a mistake wrt passing previous layer outputs to the Lambda layer? Like I have used a variable 'x' in the Lambda function assuming the previous layer output will be captured in this variable. Is this the correct way of adding the Lambda layer? (I am a beginner in Keras, so will be really grateful for any help)

deep-learning keras computer-vision

share|improve this question

edited Mar 20 at 6:35

psj

asked Mar 19 at 13:50

psjpsj

62

New contributor

psj is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  Making a separate custom layer rather than using a Lambda layer solved the issue for me. The possible reason according to me might be the use of 'tf.spectral.irfft' function inside the Lambda function. This function changes the output shape which needs to be specified using compute_output_shape.
  $endgroup$
  – psj
  2 days ago
  

  
  
  
  

add a comment | 

$begingroup$

I have the following part of code -

...
model.add(Generic_act_func())
print('layer 1 output')
print(model.layers[1].output) #Here layers[1] is Generic_act_func() layer specified above
model.add(Lambda(lambda x: tf.spectral.irfft2d(x)))
...

Which when run produces the following output and error respectively -

Output -

layer 1 output
Tensor("generic_act_func_1/Complex:0", shape=(?, 28, 28, 9), dtype=complex64)

Error -

ValueError: Tensor conversion requested dtype complex64 for Tensor with dtype float32: 'Tensor("lambda_1/Placeholder:0", shape=(?, 28, 28, 9), dtype=float32)'
at line

model.add(Lambda(lambda x: tf.spectral.irfft2d(x)))

tf.spectral.irfft2d() requires a complex64 input which the previous layer i.e., Generic_act_func() layer is already producing (as printed by layer 1 output), hence I am not able to figure out why this error. Thanks in advance for any help for solving the error.

Edit - Could there be a mistake wrt passing previous layer outputs to the Lambda layer? Like I have used a variable 'x' in the Lambda function assuming the previous layer output will be captured in this variable. Is this the correct way of adding the Lambda layer? (I am a beginner in Keras, so will be really grateful for any help)

deep-learning keras computer-vision

share|improve this question

edited Mar 20 at 6:35

psj

asked Mar 19 at 13:50

psjpsj

62

New contributor

psj is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

$endgroup$

...
model.add(Generic_act_func())
print('layer 1 output')
print(model.layers[1].output) #Here layers[1] is Generic_act_func() layer specified above
model.add(Lambda(lambda x: tf.spectral.irfft2d(x)))
...

layer 1 output
Tensor("generic_act_func_1/Complex:0", shape=(?, 28, 28, 9), dtype=complex64)

ValueError: Tensor conversion requested dtype complex64 for Tensor with dtype float32: 'Tensor("lambda_1/Placeholder:0", shape=(?, 28, 28, 9), dtype=float32)'
at line

model.add(Lambda(lambda x: tf.spectral.irfft2d(x)))

share|improve this question

edited Mar 20 at 6:35

psj

asked Mar 19 at 13:50

psjpsj

62

New contributor

psj is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

share|improve this question

edited Mar 20 at 6:35

psj

asked Mar 19 at 13:50

psjpsj

62

New contributor

psj is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

asked Mar 19 at 13:50

psjpsj

62

New contributor

psj is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

asked Mar 19 at 13:50

psjpsj

62