https://community.databricks.com/t5/data-engineering/when-can-horovod-be-used-for-an-ml-problem/m-p/25945#M18113 <P>When can Horovod be used for an ML problem?</P> Mon, 07 Jun 2021 17:59:10 GMT User16788317466 2021-06-07T17:59:10Z https://community.databricks.com/t5/data-engineering/when-can-horovod-be-used-for-an-ml-problem/m-p/25945#M18113 <P>When can Horovod be used for an ML problem?</P> Mon, 07 Jun 2021 17:59:10 GMT https://community.databricks.com/t5/data-engineering/when-can-horovod-be-used-for-an-ml-problem/m-p/25945#M18113 User16788317466 2021-06-07T17:59:10Z https://community.databricks.com/t5/data-engineering/when-can-horovod-be-used-for-an-ml-problem/m-p/25946#M18114 <P>Only when you have a gradient-descent problem. Pytorch and Tensorflow are the only candidate frameworks to use here. When using Horovod, start with single node, multi-GPU and measure training performance. If this is not sufficient, look at a multi-node, multi-GPU setup.</P> Mon, 07 Jun 2021 18:02:24 GMT https://community.databricks.com/t5/data-engineering/when-can-horovod-be-used-for-an-ml-problem/m-p/25946#M18114 User16788317466 2021-06-07T18:02:24Z