Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

implement numOutputs for biasAdd #129

Open
silky opened this issue Jun 5, 2017 · 3 comments
Open

implement numOutputs for biasAdd #129

silky opened this issue Jun 5, 2017 · 3 comments

Comments

@silky
Copy link
Contributor

silky commented Jun 5, 2017

this function fails with:

numOuputs not implemented for "BiasAdd"

when you try and use it anywhere, such as

...
conv `TFC.biasAdd` (TF.readValue convBias)
@fkm3
Copy link
Contributor

fkm3 commented Jun 5, 2017

I need to improve that error message. It just means that the gradients aren't implemented for BiasAdd. There need to be implementations for both numOutputs and opGrad.

@silky
Copy link
Contributor Author

silky commented Jun 5, 2017

ah yes, of course

@fkm3
Copy link
Contributor

fkm3 commented Jun 5, 2017

If you want biasAdd because of conv2d with NCHW format, then, coincidentally, I'm doing something similar, and I think you can work around the lack of biasAdd by using normal add and putting some 1's on the end of your bias shape, e.g. [numChannels, 1, 1].

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants