model structure

[cmf1997]

great work of scFAN!
hi, I was just getting on deep learning,
I have a question about the conv layer in the model
the input size is 1,1000,6, in my understanding the 1 the extend dim for conv2d,
and the size of kernel is (1,8), which is larger than 6
can you explain what is the advantage for tf binding prediction?

any suggestion would be much appreciated !

[cmf1997]

and the size of kernel is (1,8), which is larger than 6
which mean the conv is conducting the the dim having 6 feature