I wanted few opinions from community members regarding rasa model training. Where do you train your model before pushing it to Raxa X cluster.
- Inside Rasa Worker Container
- Outside Rasa X - GPU based server
- Outside Rasa X - CPU based server
- Others (I will reply in comment)
I’ve found that it’s kind of infeasible to train your model in Rasa X once it gets to a certain size. Our model takes around 4-5 hours in comparison. It’s much better to train it on a separate server then upload it to Rasa X after.
As for the CPU vs GPU server, I can’t seem to find a definite answer to this. It also comes down to cost vs time.