Is using Booster.predict() thread safe in scala?

#1

I’ve trained a model in python and would like to server it on a high throughput webserver. Is it thread safe?

Would it work in parallel? Or would it lock?

#2

It is thread safe, but it uses a lock, so your concurrent requests will be serialized.

We recently added capacity to support in-place prediction in #5389 but it is currently only available from Python.