Tensor.Serialization.ixx File Reference

Tensor-specific serialization helpers and metadata. More...

#include <string>
#include <cstddef>
#include <stdexcept>
#include <utility>
import Compute.CpuMemoryResource;
import Dnn.TensorBuffer;
import Dnn.TensorTypes;
import Dnn.TensorDataTypeTraits;
import Dnn.TensorDataType;
import Serialization.Metadata;
import Serialization.ModelArchive;

Classes
struct	Mila::Dnn::Serialization::ITensorBlob
	Type-erased interface for a serialized tensor blob. More...
struct	Mila::Dnn::Serialization::TensorBlob< MR >
	Concrete tensor blob owning a TensorBuffer-backed raw byte buffer. More...
struct	Mila::Dnn::Serialization::TensorBlobView
	Non-owning ITensorBlob view over externally-owned bytes. More...
struct	Mila::Dnn::Serialization::TensorMetadata
	Metadata describing a tensor in serialized form. More...

Namespaces
namespace	Mila
	Mila main API namespace.

Functions
template<typename MR = Compute::CpuMemoryResource> requires isValidTensor<dtype_t::UINT8, MR>
TensorBlob< MR >	Mila::Dnn::Serialization::readTensorBlob (const ModelArchive &archive, const std::string &prefix, int device_id=0)
	Read tensor metadata and raw bytes from prefix into a typed blob.
void	Mila::Dnn::Serialization::writeTensorBlob (ModelArchive &archive, const std::string &prefix, const TensorMetadata &meta, const void *data, size_t size)
	Write tensor metadata and raw bytes under the given prefix into archive.

Detailed Description

Tensor-specific serialization helpers and metadata.

Provides TensorMetadata, ITensorBlob, and free helpers to write/read tensor metadata + raw bytes using ModelArchive. This partition keeps tensor concerns out of ModelArchive and lets tensor implementations call these helpers.

Function Documentation

readTensorBlob()

template<typename MR = Compute::CpuMemoryResource>
requires isValidTensor<dtype_t::UINT8, MR>

TensorBlob< MR > Mila::Dnn::Serialization::readTensorBlob ( const ModelArchive & archive, const std::string & prefix, int device_id = 0 )

export

Read tensor metadata and raw bytes from prefix into a typed blob.

Allocates a TensorBuffer<UINT8, MR> sized to total_bytes and reads directly into it via readBlobInto – no intermediate vector copy. When MR is CudaPinnedMemoryResource the returned blob carries a pinned host pointer ready for direct DMA in copyFromBlob.

Template Parameters

MR	Memory resource for the blob data buffer. Defaults to CpuMemoryResource.

Parameters

archive	ModelArchive to read from
prefix	Path prefix for tensor files (e.g., "tensors/weight")
device_id	Device index passed to the memory resource constructor.

Returns

TensorBlob<MR> owning the metadata and raw byte buffer.

Exceptions

std::runtime_error

if read operations fail or size mismatch detected

writeTensorBlob()

void Mila::Dnn::Serialization::writeTensorBlob ( ModelArchive & archive, const std::string & prefix, const TensorMetadata & meta, const void * data, size_t size )

inlineexport

Write tensor metadata and raw bytes under the given prefix into archive.

Writes: prefix + "/meta.json" – TensorMetadata as SerializationMetadata prefix + "/data.bin" – raw tensor bytes

Parameters

archive	ModelArchive to write to
prefix	Path prefix for tensor files (e.g., "tensors/weight")
meta	Tensor metadata
data	Pointer to raw tensor bytes
size	Number of bytes to write

Exceptions

std::runtime_error

if write operations fail