From issues-return-6982-archive-asf-public=cust-asf.ponee.io@mxnet.apache.org Thu Oct 15 15:57:55 2020 Return-Path: X-Original-To: archive-asf-public@cust-asf.ponee.io Delivered-To: archive-asf-public@cust-asf.ponee.io Received: from mxout1-he-de.apache.org (mxout1-he-de.apache.org [95.216.194.37]) by mx-eu-01.ponee.io (Postfix) with ESMTPS id 29E2E180654 for ; Thu, 15 Oct 2020 17:57:55 +0200 (CEST) Received: from mail.apache.org (mailroute1-lw-us.apache.org [207.244.88.153]) by mxout1-he-de.apache.org (ASF Mail Server at mxout1-he-de.apache.org) with SMTP id 9D54262CE4 for ; Thu, 15 Oct 2020 15:57:54 +0000 (UTC) Received: (qmail 40114 invoked by uid 500); 15 Oct 2020 15:57:54 -0000 Mailing-List: contact issues-help@mxnet.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: dev@mxnet.apache.org Delivered-To: mailing list issues@mxnet.apache.org Received: (qmail 40105 invoked by uid 99); 15 Oct 2020 15:57:53 -0000 Received: from ec2-52-202-80-70.compute-1.amazonaws.com (HELO gitbox.apache.org) (52.202.80.70) by apache.org (qpsmtpd/0.29) with ESMTP; Thu, 15 Oct 2020 15:57:53 +0000 From: =?utf-8?q?GitBox?= To: issues@mxnet.apache.org Subject: =?utf-8?q?=5BGitHub=5D_=5Bincubator-mxnet=5D_zjost_commented_on_issue_=23193?= =?utf-8?q?23=3A_Variable_sequence_length_not_handled_correctly_for_BiDirect?= =?utf-8?q?ional_layers?= Message-ID: <160277747387.32230.11179398767440784862.asfpy@gitbox.apache.org> Date: Thu, 15 Oct 2020 15:57:53 -0000 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit In-Reply-To: References: zjost commented on issue #19323: URL: https://github.com/apache/incubator-mxnet/issues/19323#issuecomment-709422989 Thanks for taking the time to show an implementation of this. Do you think a warning should be added to the documentation regarding the use of `bidirectional`? I don't think it's clear that this will not have intended behavior unless sequences are of equal length in a batch. Particularly so because if you start tracing down the base classes, it seems as though e.g. GRU will pass along the `use_sequence_length` kwarg. Regarding #14208, I'm not sure why this fails for other RNN types since the code changes appear to be primarily to the `_RNN` base class rather than the LSTMCell. ---------------------------------------------------------------- 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: users@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscribe@mxnet.apache.org For additional commands, e-mail: issues-help@mxnet.apache.org