Hi guys, just wanted to highlight that currently the bow.py example will fail when not GPUs are available due to the hard-coded .cuda conversions in senteval/tools/classifier.py. An easy fix is to init the classifier with a device variable, and simply set .to(self.device) everywhere. Thanks for making this framework available!
Hi guys, just wanted to highlight that currently the
bow.py
example will fail when not GPUs are available due to the hard-coded.cuda
conversions insenteval/tools/classifier.py
. An easy fix is to init the classifier with adevice
variable, and simply set.to(self.device)
everywhere. Thanks for making this framework available!