There is an optimizer in ClearML already.
Here is an example:
and some docs 🙂
I would also suggest using pipelines if you want to do several actions with a task controlling the progress.
Alright! I'll take a look at it. It's also nice to know that pipelines will take care of it. Thanks!
Hmm okay, I'm doing a hyper parameter search by launching multiple processes of my
train function. I've got a main task runing the search to log the final results, and a bunch of training tasks running in parallel. It would've been nice to be able to come back to each one individual training task, but I guess I'll do without
you're always running a single task at a time. The whole point is that everything is reported to the task (auto-magic bindings, console logs etc.), so there cannot be any ambiguity. You can close the current task (
task.close() ) and init a new one if you'd like, but you can't init several at the same time.