Can't train (failed to connect to kubernetes.docker.internal and unsupported gloo device)

[magicnothief]

Describe the bug
Training can't begin because of an error.

To Reproduce
Steps to reproduce the behavior:

1. Go to Train
2. Click on train model
3. See error

Expected behavior
An error mentioning kubernetes.docker.internal and unsupported gloo device

Screenshots

Desktop (please complete the following information):

• OS and version: Windows
• Python version: 3.11

[fumiama]

Currently train only support running on NVIDIA GPU.

[magicnothief]

I have an NVIDIA RTX 3060.