Keras: Access true output tensor within model, possible?

[carlthome]

Is it possible to define a model that uses the values of the target tensors in some layers? I don't mean the current output predictions, but the actual reference target values.

[massivezzoli]

I am looking for an answer to this question too. any luck on your side?

[fchollet]

Make your targets an input to the model. But then of course, during
inference you will still need to pass something as "targets".

On 4 April 2017 at 04:13, massivezzoli [e-mail] wrote:

I am looking for an answer to this question too. any luck on your side?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
https://github.com/fchollet/keras/issues/2859#issuecomment-291374527,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AArWb33RCANCUC51p3OV8rlIznc1Zwq9ks5rsadfgaJpZM4IqhPQ
.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed after 30 days if no further activity occurs, but feel free to re-open a closed issue if needed.

[sirius0503]

Make your targets an input to the model. But then of course, during inference you will still need to pass something as "targets".
…
On 4 April 2017 at 04:13, massivezzoli @.*> wrote: I am looking for an answer to this question too. any luck on your side? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2859 (comment)>, or mute the thread https://github.com/notifications/unsubscribe-auth/AArWb33RCANCUC51p3OV8rlIznc1Zwq9ks5rsadfgaJpZM4IqhPQ .

@fchollet : Can you explain this better, at the moment, I am getting only the tensor values, not the numpy values and K.eval doesn't work asking me to feed it feed_dict, which would require the images again.