5cacecc0c0
* [CUBLAS] Add support for nn.dense and nn.batch_matmul This commit includes a fix for cublas.batch_matmul when mixed precision is being used. * Specify args in dense