Can Rasa do distributed training?

Can Rasa do distributed training? For example, the layout is on k8s

Yep, I think so

Any hints on that process? So far I didn’t manage to distribute my NLU training over multiple GPUs.

distributed training is not supported at the moment

1 Like

@Ghostvv Is it possible to do distributed training in rasa version 2.8.6 or Are we supporting in any other version now?

I would also like to know that. My idea is to use Ray cluster for multi-node distributed training, with each node representing a separate HW, having a Rasa chatbot of its own.