Hello, Could anybody suggest any work around for an issue with nan gradients occurred in T.switch if one of its branch produces nan? https://github.com/Theano/Theano/issues/5333 . Seems like it's the same as this problem https://github.com/Theano/Theano/issues/4918 but the issue creator claimed that T.ifelse didn't work for him, but T.switch did.
Best Regards, Minh -- --- You received this message because you are subscribed to the Google Groups "theano-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
