shanyang-me

Hi all, I found it confusing when I look at Figure 11 on page 15. It uses a MLP (2 layers) and output dim is 64, but if it's predicting...