提交 507f4797 编写于 作者: M minqiyang

Polish code

上级 d49ff1a8
......@@ -1111,7 +1111,7 @@ def multi_box_head(inputs,
mbox_loc = nn.transpose(mbox_loc, perm=[0, 2, 3, 1])
compile_shape = [
mbox_loc.shape[0], cpt.floor_division(
box_loc.shape[1] * mbox_loc.shape[2] * mbox_loc.shape[3], 4), 4
mbox_loc.shape[1] * mbox_loc.shape[2] * mbox_loc.shape[3], 4), 4
]
run_shape = tensor.assign(numpy.array([0, -1, 4]).astype("int32"))
mbox_loc_flatten = nn.reshape(
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册