swfearp commented on issue #17881: Solved Bug: contrib/amp/loss_scalar.py not initializing variables URL: https://github.com/apache/incubator-mxnet/issues/17881#issuecomment-607702769 Hey, here are the amp lines: ``` amp.init() ``` ... optimizer = mx.optimizer.SGD(learning_rate=self.lr, momentum=self.momentum, wd=self.wd) trainer = gluon.Trainer(params=self.model.collect_params(), optimizer=optimizer) amp.init_trainer(trainer) ``` ``` with autograd.record(): outputs = [self.model(X) for X in data] losses = [loss_function(output, label) for output, label in zip(outputs, smooth_labels)] with amp.scale_loss(losses, trainer) as scaled_loss: autograd.backward(scaled_loss) ```
---------------------------------------------------------------- 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
