GptBlock.ixx File Reference

Transformer encoder block implementation. More...

#include <memory>
#include <vector>
#include <string>
#include <sstream>
#include <iostream>
#include <iomanip>
#include <format>
#include <utility>
#include <stdexcept>
#include <cstdint>
#include <type_traits>
#include <optional>
#include <algorithm>
import Logging.Logger;
import Dnn.CompositeComponent;
import Dnn.Components.MultiHeadAttention;
import Compute.MemoryResource;
import Dnn.Components.LayerNorm;
import Dnn.Component;
import Dnn.Components.Residual;
import Dnn.TensorOps;
import Dnn.TensorDataTypeTraits;
import Dnn.ActivationType;
import Dnn.TensorDataType;
import Compute.ExecutionContextFactory;
import Dnn.TensorTypes;
import Compute.CpuMemoryResource;
import Dnn.Components.Linear;
import Dnn.ITensor;
import Compute.IExecutionContext;
import Dnn.TensorInitializers;
import Dnn.ComponentType;
import Dnn.TensorHelpers;
import Dnn.Tensor;
import Dnn.Components.MLP;
import Compute.DeviceId;
import Dnn.Components.GptBlock:Config;
import Compute.DeviceType;
import Compute.DeviceTypeTraits;
import Compute.Device;
import Serialization.ModelArchive;
import Compute.ExecutionContext;
import Serialization.Mode;

Classes
class	Mila::Dnn::GptBlock< TDeviceType, TPrecision >
	Transformer encoder block as a composite component. More...

Namespaces
namespace	Mila
	Mila main API namespace.
namespace	Mila::Dnn

Detailed Description

Transformer encoder block implementation.

Provides a device-templated composite Transformer encoder block that composes attention, layer-norm, residual connections and MLP sub-components.