Serializers

Implements base methods for serializing data for an inference endpoint.

class sagemaker.base_serializers.BaseSerializer

Bases: ABC

Abstract base class for creation of new serializers.

Provides a skeleton for customization requiring the overriding of the method serialize and the class attribute CONTENT_TYPE.

abstract serialize(data)

Serialize data into the media type specified by CONTENT_TYPE.

Parameters:: data (object) – Data to be serialized.
Returns:: Serialized data used for a request.
Return type:: object

abstract property CONTENT_TYPE: The MIME type of the data sent to the inference endpoint.

class sagemaker.base_serializers.SimpleBaseSerializer(content_type='application/json')

Bases: BaseSerializer

Abstract base class for creation of new serializers.

This class extends the API of :class:~`sagemaker.serializers.BaseSerializer` with more user-friendly options for setting the Content-Type header, in situations where it can be provided at init and freely updated.

Initialize a SimpleBaseSerializer instance.

Parameters:

content_type (str) – The MIME type to signal to the inference endpoint when sending
(default (request data) – “application/json”).

property CONTENT_TYPE: The data MIME type set in the Content-Type header on prediction endpoint requests.

class sagemaker.base_serializers.CSVSerializer(content_type='text/csv')

Bases: SimpleBaseSerializer

Serialize data of various formats to a CSV-formatted string.

Initialize a CSVSerializer instance.

Parameters:: content_type (str) – The MIME type to signal to the inference endpoint when sending request data (default: “text/csv”).

serialize(data)

Serialize data of various formats to a CSV-formatted string.

Parameters:: data (object) – Data to be serialized. Can be a NumPy array, list, file, Pandas DataFrame, or buffer.
Returns:: The data serialized as a CSV-formatted string.
Return type:: str

class sagemaker.base_serializers.NumpySerializer(dtype=None, content_type='application/x-npy')

Bases: SimpleBaseSerializer

Serialize data to a buffer using the .npy format.

Initialize a NumpySerializer instance.

Parameters:

content_type (str) – The MIME type to signal to the inference endpoint when sending request data (default: “application/x-npy”).
dtype (str) – The dtype of the data.

serialize(data)

Serialize data to a buffer using the .npy format.

Parameters:: data (object) – Data to be serialized. Can be a NumPy array, list, file, or buffer.
Returns:: A buffer containing data serialzied in the .npy format.
Return type:: io.BytesIO

class sagemaker.base_serializers.JSONSerializer(content_type='application/json')

Bases: SimpleBaseSerializer

Serialize data to a JSON formatted string.

Initialize a SimpleBaseSerializer instance.

Parameters:

content_type (str) – The MIME type to signal to the inference endpoint when sending
(default (request data) – “application/json”).

serialize(data)

Serialize data of various formats to a JSON formatted string.

Parameters:: data (object) – Data to be serialized.
Returns:: The data serialized as a JSON string.
Return type:: str

class sagemaker.base_serializers.IdentitySerializer(content_type='application/octet-stream')

Bases: SimpleBaseSerializer

Serialize data by returning data without modification.

This serializer may be useful if, for example, you’re sending raw bytes such as from an image file’s .read() method.

Initialize an IdentitySerializer instance.

Parameters:: content_type (str) – The MIME type to signal to the inference endpoint when sending request data (default: “application/octet-stream”).

serialize(data)

Return data without modification.

Parameters:: data (object) – Data to be serialized.
Returns:: The unmodified data.
Return type:: object

class sagemaker.base_serializers.JSONLinesSerializer(content_type='application/jsonlines')

Bases: SimpleBaseSerializer

Serialize data to a JSON Lines formatted string.

Initialize a JSONLinesSerializer instance.

Parameters:: content_type (str) – The MIME type to signal to the inference endpoint when sending request data (default: “application/jsonlines”).

serialize(data)

Serialize data of various formats to a JSON Lines formatted string.

Parameters:

data (object) – Data to be serialized. The data can be a string, iterable of JSON serializable objects, or a file-like object.

Returns:

The data serialized as a string containing newline-separated: JSON values.

Return type:

str

class sagemaker.base_serializers.SparseMatrixSerializer(content_type='application/x-npz')

Bases: SimpleBaseSerializer

Serialize a sparse matrix to a buffer using the .npz format.

Initialize a SparseMatrixSerializer instance.

Parameters:: content_type (str) – The MIME type to signal to the inference endpoint when sending request data (default: “application/x-npz”).

serialize(data)

Serialize a sparse matrix to a buffer using the .npz format.

Sparse matrices can be in the csc, csr, bsr, dia or coo formats.

Parameters:: data (scipy.sparse.spmatrix) – The sparse matrix to serialize.
Returns:: A buffer containing the serialized sparse matrix.
Return type:: io.BytesIO

class sagemaker.base_serializers.LibSVMSerializer(content_type='text/libsvm')

Bases: SimpleBaseSerializer

Serialize data of various formats to a LibSVM-formatted string.

The data must already be in LIBSVM file format: <label> <index1>:<value1> <index2>:<value2> …

It is suitable for sparse datasets since it does not store zero-valued features.

Initialize a LibSVMSerializer instance.

Parameters:: content_type (str) – The MIME type to signal to the inference endpoint when sending request data (default: “text/libsvm”).

serialize(data)

Serialize data of various formats to a LibSVM-formatted string.

Parameters:: data (object) – Data to be serialized. Can be a string or a file-like object.
Returns:: The data serialized as a LibSVM-formatted string.
Return type:: str
Raises:: ValueError – If unable to handle input format

class sagemaker.base_serializers.DataSerializer(content_type='file-path/raw-bytes')

Bases: SimpleBaseSerializer

Serialize data in any file by extracting raw bytes from the file.

Initialize a DataSerializer instance.

Parameters:: content_type (str) – The MIME type to signal to the inference endpoint when sending request data (default: “file-path/raw-bytes”).

serialize(data)

Serialize file data to a raw bytes.

Parameters:: data (object) – Data to be serialized. The data can be a string representing file-path or the raw bytes from a file.
Returns:: The data serialized as raw-bytes from the input.
Return type:: raw-bytes

class sagemaker.base_serializers.StringSerializer(content_type='text/plain')

Bases: SimpleBaseSerializer

Encode the string to utf-8 bytes.

Initialize a StringSerializer instance.

Parameters:: content_type (str) – The MIME type to signal to the inference endpoint when sending request data (default: “text/plain”).

serialize(data)

Encode the string to utf-8 bytes.

Parameters:: data (object) – Data to be serialized.
Returns:: The data serialized as raw-bytes from the input.
Return type:: raw-bytes

class sagemaker.base_serializers.TorchTensorSerializer(content_type='tensor/pt')

Bases: SimpleBaseSerializer

Serialize torch.Tensor to a buffer by converting tensor to numpy and call NumpySerializer.

Parameters:: data (object) – Data to be serialized. The data must be of torch.Tensor type.
Returns:: The data serialized as raw-bytes from the input.
Return type:: raw-bytes

Initialize a SimpleBaseSerializer instance.

Parameters:

content_type (str) – The MIME type to signal to the inference endpoint when sending
(default (request data) – “application/json”).

serialize(data)

Serialize torch.Tensor to a buffer.

Parameters:: data (object) – Data to be serialized. The data must be of torch.Tensor type.
Returns:: The data serialized as raw-bytes from the input.
Return type:: raw-bytes

class sagemaker.base_serializers.RecordSerializer(content_type='application/x-recordio-protobuf')

Bases: SimpleBaseSerializer

Serialize a NumPy array for an inference request.

Initialize a RecordSerializer instance.

Parameters:: content_type (str) – The MIME type to signal to the inference endpoint when sending request data (default: “application/x-recordio-protobuf”).

serialize(data)

Serialize a NumPy array into a buffer containing RecordIO records.

Parameters:: data (numpy.ndarray) – The data to serialize.
Returns:: A buffer containing the data serialized as records.
Return type:: io.BytesIO

Implements methods for serializing data for an inference endpoint.

sagemaker.serializers.retrieve_options(region=None, model_id=None, model_version=None, hub_arn=None, tolerate_vulnerable_model=False, tolerate_deprecated_model=False, sagemaker_session=<sagemaker.session.Session object>, config_name=None)

Retrieves the supported serializers for the model matching the given arguments.

Parameters:

region (str) – The AWS Region for which to retrieve the supported serializers. Defaults to None.
model_id (str) – The model ID of the model for which to retrieve the supported serializers. (Default: None).
model_version (str) – The version of the model for which to retrieve the supported serializers. (Default: None).
hub_arn (str) – The arn of the SageMaker Hub for which to retrieve model details from. (Default: None).
tolerate_vulnerable_model (bool) – True if vulnerable versions of model specifications should be tolerated (exception not raised). If False, raises an exception if the script used by this version of the model has dependencies with known security vulnerabilities. (Default: False).
tolerate_deprecated_model (bool) – True if deprecated models should be tolerated (exception not raised). False if these models should raise an exception. (Default: False).
sagemaker_session (sagemaker.session.Session) – A SageMaker Session object, used for SageMaker interactions. If not specified, one is created using the default AWS configuration chain. (Default: sagemaker.jumpstart.constants.DEFAULT_JUMPSTART_SAGEMAKER_SESSION).
config_name (Optional[str]) – Name of the JumpStart Model config to apply. (Default: None).

Returns:

The supported serializers to use for the model.

Return type:

List[SimpleBaseSerializer]

Raises:

ValueError – If the combination of arguments specified is not supported.

sagemaker.serializers.retrieve_default(region=None, model_id=None, model_version=None, hub_arn=None, tolerate_vulnerable_model=False, tolerate_deprecated_model=False, model_type=JumpStartModelType.OPEN_WEIGHTS, sagemaker_session=<sagemaker.session.Session object>, config_name=None)

Retrieves the default serializer for the model matching the given arguments.

Parameters:

region (str) – The AWS Region for which to retrieve the default serializer. Defaults to None.
model_id (str) – The model ID of the model for which to retrieve the default serializer. (Default: None).
model_version (str) – The version of the model for which to retrieve the default serializer. (Default: None).
hub_arn (str) – The arn of the SageMaker Hub for which to retrieve model details from. (Default: None).
tolerate_vulnerable_model (bool) – True if vulnerable versions of model specifications should be tolerated (exception not raised). If False, raises an exception if the script used by this version of the model has dependencies with known security vulnerabilities. (Default: False).
tolerate_deprecated_model (bool) – True if deprecated models should be tolerated (exception not raised). False if these models should raise an exception. (Default: False).
sagemaker_session (sagemaker.session.Session) – A SageMaker Session object, used for SageMaker interactions. If not specified, one is created using the default AWS configuration chain. (Default: sagemaker.jumpstart.constants.DEFAULT_JUMPSTART_SAGEMAKER_SESSION).
config_name (Optional[str]) – Name of the JumpStart Model config to apply. (Default: None).
model_type (JumpStartModelType) –

Returns:

The default serializer to use for the model.

Return type:

SimpleBaseSerializer

Raises:

ValueError – If the combination of arguments specified is not supported.