Run Time Error: expected mat1 and mat2 to have same dtype while implementing GraphSage via Pytorch Geometric

While implementing GraphSage code via Pytorch Geometric, I am using the Neighbour_Loader function to do minibatching. I am working on graphdata wit size in 410 million.
Graphsage takes 3 parameters edge index, x and y. When I am running the code, I am getting the run time error expected mat1 and mat2 to have same dtype, but got long int!=float. I have not type casted any of the pytorch geometric objects or the tensors. All of them are int64. Please someone solve this error.