Comment by costco

4 days ago

This is a result of the use of positional embeddings, which typically results in the final item being weighted very highly. The problem is that this information is shown to be very relevant to the task of predicting the next item interacted with. If you add more books the effect of this is somewhat diluted.