Why Should we Combine Training and Post-Training Methods for Out-of-Distribution Detection?

Open in new window