question about the attention map visualization

Question

question about the attention map visualization

amiltonwong opened this issue 4 years ago · 2 comments

For Figure 1 in your paper, which self-attention layer is used to visualize the attention map? From your implementation, there are 4 self-attention layers (SA1, SA2, SA3, SA4) in the model.

Thanks~

Answer 1 · 2021-03-24T14:47:48.000Z

Hi, @amiltonwong
Like Vison Transformer, we visualize the attention map by using mean value of all attention layer.

Answer 2 · 2021-03-25T00:37:55.000Z

@MenghaoGuo , OK, Thanks ~