KevinMac

Follow up on this, the difference in output values was because I forgot to put the model in evaluate mode in PyTorch (stupid!). The main issue outlined here is still...

Hey @Lothiraldan, can do. It's a pretty simple use case. I just occasionally use an OfflineExperiment to store things locally if our internet is acting up or if I want...

Writing to bump this. I notice the labels are actually missing for the plots on the web interface as well, this makes them nearly impossible to read, especially if you...

Thanks for looking into this. > If you want to train each qa pair as an example and pad to a fixed length or use datacollator to pad to longest...