Kurt Shuster

Results 198 comments of Kurt Shuster

Hi there, yes you've identified a mismatch between training and test-time inference. The memory decision task during training only shows a single persona; this is meant to teach the model...

oops, that should be `http://parl.ai/downloads/_models/bb3/bb3_30B/consolidated.pt`. let me edit the README

oh the README is up to date, that is the correct URL ^

is this a warning, or does your train script exit unsuccessfully?

Ok, it should not affect any of your training scripts if it's just a warning

What you've provided should work

That is the correct way to do it. BB3 should stay on topic most of the time