ptrendx commented on issue #19367: URL: https://github.com/apache/incubator-mxnet/issues/19367#issuecomment-712393379
You can use `LARS` optimizer - it uses the same kernels as the MLPerf submission (they are named slightly differently - ` preloaded_multi_mp_sgd_update`). The NVIDIA NGC MXNet container, which was used for that submission, had the LARS optimizer before the upstream and the names were changed as part of the upstreaming process. ---------------------------------------------------------------- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
