GPyOpt producing real values for discrete parameters
AlexFuster opened this issue · 2 comments
AlexFuster commented
Running the example mnist_mlp, GPyOpt gave me the following configuration:
Trial 4: {'num_units': 116.0, 'activation': 'relu'}
This obviously makes Keras to produce an error, as it does not admit real values for number of units
I know this is probably due to the changes made to solve the issue #42
LarsHH commented
Nice catch! Thanks. Yeah this was part of me changing the conversion from
Sherpa parameters to GPyOpt Parameters and the tests didn’t catch it. I’ll
fix it later today.
…On Mon, Jun 3, 2019 at 6:26 AM AlexFuster ***@***.***> wrote:
Running the example mnist_mlp, GPyOpt gave me the following configuration:
Trial 4: {'num_units': 116.0, 'activation': 'relu'}
This obviously makes Keras to produce an error, as it does not admit real
values for number of units
I know this is probably due to the changes made to solve the issue #42
<#42>
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#48?email_source=notifications&email_token=ABZVYYU5DGGYMJDV33VACM3PYULWXA5CNFSM4HSIBE22YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4GXJCSGA>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABZVYYUNNWKVE46WM6SYS63PYULWXANCNFSM4HSIBE2Q>
.