StevenJokes edited a comment on issue #18931: URL: https://github.com/apache/incubator-mxnet/issues/18931#issuecomment-676554660
And one more thing, the new d2l [forum](https://discuss.d2l.ai/) have no offical reply three days. I'm a newbie. I'm trying to answer some, but I'm afraid to mislead others. Please spare some time to answer some if you guys can. --- I'm thinking that the forum is most important to get more new users and find pain spots when communicating. The pytorch's friendly is caused by the saved discussions. But if we use our forum correctly, we will beat them. --- Our discussion is just after the chapters' context. It will be more systemic. And we should rule the dicussions to depulicate less. My opinion here: We must let others to know that you cared about them when using our framework. And the efforts about code can be delayed a little, because we need to train the first wave users from tf and pytorch to let them get familiar to mxnet. Then they can train the next and next waves. I think one reason why mxnet 1.0 failed is that we don't care others and give no reply to their questions in discuss.mxnet.io . We need to discuss more mxnet details other than just throw a universal code or docs. --- The old developers must spare your precious time to test all mxnet code and fix them ASAP. The old developers must spare your precious time to answer questions ASAP when we adding pytorch and tensorflow context, to attract other tf and pytorch users to coming. **And we should give them reasons why our framework is good after finishing replying.** --- d2l is the our only book, and we can find at least 5 books of pytorch and tensorflow. If we don't grasp this chance, d2l will just help pytorch and tf developing and make mxnet more deadly. All my thoughts. ---------------------------------------------------------------- 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]
