attention_matrices, # list of 2d tensors T1xT2, T2xT3, ..., Tk-1xTk
# do not draw links with a lesser attention
min_link_attention=0,
- # draw only the strongest links necessary to reache
- # min_total_attention
+ # draw only the strongest links necessary so that their summed
+ # attention is above min_total_attention
min_total_attention=None,
# draw only the top k links
k_top=None,