[Tune] Ray tune terminates after OSError: [Errno 28] No space left on device

[davelbit]

Hello everyone,

I am new to using ray tune for hyperparameter tuning and therefore not that experienced in using it. However, I followed the PyTorch tutorial (https://docs.ray.io/en/master/tune/tutorials/tune-tutorial.html#tutorial-tune-setup) so that I can optimize my hyperparameters, but after 10 trials my system crashes because apparently there is no more memory (disk space) available. And my question now is, how can I use RayTune so that I can finish my hyperparameter optimization without worrying about my disk space.

FYI: num_sample = 20, epochs=20

[rliaw]

Hey @davelbit,

This is a bug that we have in the 1.3 release. Can you follow the instructions here: Ray tune log grows extremely large - #3 by daniel-pp to address this for now?

[davelbit]

Thanks for the quick reply! That helps a lot.