Closed luiscarm9 closed 10 months ago
Thanks for raising this. The save_checkpoints
argument is used when using the Tensorflow backend:
Although the argument is not used in PyTorch. Rather than remove the argument, let's add feature parity for the PyTorch backend. I'll put out a patch & test in the next couple of days, and include the update in the 2.1.1 release.
Closed, as a fix was implemented in 0a0ce26
The keyword argument "save_checkpoints" set up in the method smac_search is unknown to the training method. If this functionality is intended I think it also needs to be implemented then.