Mila
Deep Neural Network Library
|
Implementation of fused matrix multiplication, bias addition, and activation operations. More...
#include <memory>
#include <vector>
#include <stdexcept>
#include <cuda_fp16.h>
#include <cuda_runtime.h>
#include <cublasLt.h>
import Compute.CudaMemoryResource;
import Compute.CudaDevice;
import Compute.DeviceContext;
import Compute.OperationType;
import Compute.OperationRegistry;
import Compute.DeviceType;
import Compute.OperationBase;
import Compute.OperationAttributes;
import Dnn.TensorTraits;
import Compute.UnaryOperation;
import Dnn.Tensor;
Classes | |
class | Mila::Dnn::Compute::CudaMatMulBiasGeluOp< TInput, TOutput > |
CUDA implementation of the fused MatMul-Bias-GELU operation. More... | |
class | Mila::Dnn::Compute::CudaMatMulBiasGeluOpRegistrar |
Class responsible for registering the CudaMatMulBiasGeluOp operation. More... | |
Namespaces | |
namespace | Mila |
namespace | Mila::Dnn |
namespace | Mila::Dnn::Compute |
Implementation of fused matrix multiplication, bias addition, and activation operations.