reminisce commented on issue #14661: [numpy] Support zero-dim and zero-size tensors in MXNet URL: https://github.com/apache/incubator-mxnet/pull/14661#issuecomment-483437787 @anirudh2290 That's because np compat scope is a thread-local variable. You need to scope the forward function where numpy compatibility is expected. Scoping the main thread does not spread the compatibility to the worker thread. The following change should work. ```python @mx.use_np_compat def forward(self, is_train, req, in_data, out_data, aux): ```
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
