A question about the implemention of ghost-pan

[imaginistLi]

I find a strange code fragment of nanodet/model/fpn/ghost_pan.py from line 215 to 218.

feat_heigh = inner_outs[0]
feat_low = inputs[idx - 1]

inner_outs[0] = feat_heigh

What dose this mean?
feat_heigh = inner_outs[0] inner_outs[0] = feat_heigh