This contribution reports the cross-check results for JVET-AF0071 where LOP2.3 filter alchitecture is modified by changing the input and output of the skip connection of Residual Block is and fusion of two adjacent 1x1 convolutions. This skip-fusion design tested in JVET-AF0071 reportedly reduces the complexity of the unified LOP2.3 from 17.05 kMac/Pixel to 16.56 kMac/Pixel and reduce the number of convolutional layers by 13%. In float implementation, the method reportedly provides BD-Rate change of {0.00%, 0.71%, 0.99%} under AI and {-0.09%, 1.55%, 1.49%} under RA.