Keras: Add support for bayesian neural netwoks with gelato (need symbolic initializer)

Created on 9 May 2017  路  4Comments  路  Source: keras-team/keras

Hi, I'm developing Gelato and in this issue I found that Keras does not support such initialization as Lasagne does. Is it possible to add support for symbolic initialization in keras?

stale

Most helpful comment

@fchollet How difficult do you think it would be to make that change? Which places would need to change?

All 4 comments

I was having the same problem until I downgraded to keras 1.2. It would be nice if this worked in 2.0

@fchollet How difficult do you think it would be to make that change? Which places would need to change?

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.

@fchollet There are several issues/requests for this on github and stackoverflow, but there's been no response from the Keras team. Is support likely to be added for symbolic initialization in coming versions of Keras?

Was this page helpful?
0 / 5 - 0 ratings