mxnet docker gives different inference output
Description
When using official mxnet docker image on cloud server and without docker on a local server, the inference result is different. The input is exactly the same.
Environment info (Required)
docker on cloud server: cuda9.0, cuda drive 418 (host machine), mxnet 1.4.0 local server: cuda9.0, cuda drive 396, mxnet 1.3.0
Hey, this is the MXNet Label Bot. Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it. Here are my recommended label(s): Performance
After downgrading the mxnet version in the docker to 1.3.0, the inference result are exactly the same. But why?
Any ideas guys?