How to debug into trainable

Monitoring & Debugging
1

hello, i tried debug into the trainable function, but it cannot enter， Does there are any other methods available?

def trainable(config, metric): 
    ...
    estimator.fit()
    ...
2

@zmin1217 By debug do you mean step-by debugger in an IDE? There is way to step into
a Ray remote function. Check the docs here.

Not sure if you can step o into a trainable, which is invoked from within a Trainer class.

cc: @xwjiang2010 any hints how to debug a Trainer’s trainable function?

3

@Jules_Damji Thank you for your reply. it is step-by debugger in an IDE(Pycharm). and i set local_mode=True, it works as following:

ray.init(local_mode=True)
4

@zmin1217 What happens when you follow the instructions in the docs how to use the debugger? Does it work?

5

@Jules_Damji i am not using the ray debugger. Just finding the parameter local_mode doc of ray.init() says it will be executed serially, and it works.

6

@zmin1217 Yes, brilliant! I’ll close this issue then. Cheers!