roywei commented on a change in pull request #16900: introduce gradient update handler to the base estimator URL: https://github.com/apache/incubator-mxnet/pull/16900#discussion_r350470977
########## File path: python/mxnet/gluon/contrib/estimator/estimator.py ########## @@ -295,13 +295,11 @@ def fit_batch(self, train_batch, batch_axis=0): with autograd.record(): pred = [self.net(x) for x in data] - loss = [self.loss(y_hat, y) for y_hat, y in zip(pred, label)] + loss = [self.loss(y_hat, y) / batch_size for y_hat, y in zip(pred, label)] Review comment: Estimator is using Loss from EvalMetrics, if loss is divided by batch_size here, it will be divided again in https://github.com/apache/incubator-mxnet/blob/master/python/mxnet/metric.py#L1626 May want to update MetricHandler or add options for `update` in `EvalMetric.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: us...@infra.apache.org With regards, Apache Git Services