dulvqingyunLT

> Try decreasing your num_workers. It might be too many for a single GPU which means most of the time is spent loading the data. Thanks, I tried to set...

@yingcanw > The log display is correct, which means that hugectr_backend does not support embedding cache sharing across GPUs. You can refer [hugectr-backend-framework](https://github.com/triton-inference-server/hugectr_backend/blob/main/docs/architecture.md#hugectr-backend-framework) for details. You can understand that each...

@yingcanw . Thanks. So Is there any plan to add the feature of 'embedding cache sharing across GPUs' to hugeCTR backend?

> Hi @dulvqingyunLT , thank you for your question! Currently NVT doesn't support Norm format output. But we are working on a notebook to demonstrate how to generate keyset with...