Natasha Jaques 2
TalkRL: The Reinforcement Learning Podcast
00:00
Is This Data Set Reusable?
The data that they use to train their reward model comes from a bunch of models that isn't just their RL model. So the data sets are totally reusable and maybe we can cost effectively build up these libraries of data sets that way. Sam: If you start training on this data and the model drifts away, then is that data set going to be still useful or is it could ever be used for another model?
Transcript
Play full episode
