I put two B and s in contiguous space like B s B s
And I send the ptr of the first B and s into the kernel as well as a pointer of offset tensor
I try to revise the code like
B = B + offset_ptr[0];
s = s + offset_ptr[1];
I want the kernel to run with the second B and s
But it seems that it dose not work
Can you give me some help?