ps_pytorch

ps_pytorch copied to clipboard

Should backward function be written every time I add a model

1

Then it's very time consuming

thtb

How to setup in my own cluster instead of AWS?

1

Hi wang, I want to setup ps-pytorch in my own cluster, but the guide only mentioned how to setup it in AWS. Can you give me some guidance about this?...

yanring

Why use float_64 for gradient

6

Hi wang, I'm just wondering why to convert the gradient Tensor into `float64`, I thought they might be just `float32`. And it should be more accurate than SGD required. https://github.com/hwang595/ps_pytorch/blob/89a1cfa136b957073576fae827d39ef0fb09d2fc/src/distributed_worker.py#L258

Stonesjtu