schema Namespace Reference

Module caffe2.python.schema. More...

Classes
class	_SchemaNode
class	Field
class	List
class	Metadata
class	Scalar
class	Struct

Functions
def	Map (keys, values, keys_name='keys', values_name='values', lengths_blob=None)
def	NamedTuple (name_prefix, fields)
def	Tuple (fields)
def	RawTuple (num_fields, name_prefix='field')
def	from_dtype (dtype, _outer_shape=())
def	from_column_list (col_names, col_types=None, col_blobs=None, col_metadata=None)
def	from_blob_list (schema, values)
def	as_record (value)
def	FetchRecord (blob_record, ws=None)
def	FeedRecord (blob_record, arrays, ws=None)
def	NewRecord (net, schema)
def	ConstRecord (net, array_record)
def	InitEmptyRecord (net, schema_or_record, enforce_types=False)
def	is_schema_subset (schema, original_schema)
def	equal_schemas (schema, original_schema)
def	schema_check (schema, previous=None)
def	data_type_for_dtype (dtype)
def	attach_metadata_to_scalars (field, metadata)

Variables
	logger = logging.getLogger(__name__)
string	FIELD_SEPARATOR = ':'
	FeatureSpec

Detailed Description

Module caffe2.python.schema.

Function Documentation

ConstRecord()

def schema.ConstRecord	(		net,
			array_record
	)

Given a record of arrays, returns a record of blobs,
initialized with net.Const.

Definition at line 931 of file schema.py.

FeedRecord()

def schema.FeedRecord	(		blob_record,
			arrays,
			ws = None
	)

Given a Record containing blob_references and arrays, which is either
a list of numpy arrays or a Record containing numpy arrays, feeds the
record to the current workspace.

Definition at line 881 of file schema.py.

FetchRecord()

def schema.FetchRecord	(		blob_record,
			ws = None
	)

Given a record containing BlobReferences, return a new record with same
schema, containing numpy arrays, fetched from the current active workspace.

Definition at line 862 of file schema.py.

from_blob_list()

def schema.from_blob_list	(		schema,
			values
	)

Create a schema that clones the given schema, but containing the given
list of values.

Definition at line 826 of file schema.py.

from_column_list()

def schema.from_column_list	(		col_names,
			col_types = None,
			col_blobs = None,
			col_metadata = None
	)

Given a list of names, types, and optionally values, construct a Schema.

Definition at line 781 of file schema.py.

from_dtype()

def schema.from_dtype	(		dtype,
			_outer_shape = ()
	)

Constructs a Caffe2 schema from the given numpy's dtype.

Numpy supports scalar, array-like and structured datatypes, as long as
all the shapes are fixed. This function breaks down the given dtype into
a Caffe2 schema containing `Struct` and `Scalar` types.

Fields containing byte offsets are not currently supported.

Definition at line 675 of file schema.py.

Map()

def schema.Map	(		keys,
			values,
			keys_name = 'keys',
			values_name = 'values',
			lengths_blob = None
	)

A map is a List of Struct containing keys and values fields.
Optionally, you can provide custom name for the key and value fields.

Definition at line 644 of file schema.py.

NewRecord()

def schema.NewRecord	(		net,
			schema
	)

Given a record of np.arrays, create a BlobReference for each one of them,
returning a record containing BlobReferences. The name of each returned blob
is NextScopedBlob(field_name), which guarantees unique name in the current
net. Use NameScope explicitly to avoid name conflictions between different
nets.

Definition at line 908 of file schema.py.

RawTuple()

def schema.RawTuple	(		num_fields,
			name_prefix = 'field'
	)

Creates a tuple of `num_field` untyped scalars.

Definition at line 666 of file schema.py.

Tuple()

def schema.Tuple

(

fields

)

Creates a Struct with default, sequential, field names of given types.

Definition at line 659 of file schema.py.

Variable Documentation

FeatureSpec

schema.FeatureSpec

Initial value:

=  namedtuple(
    'FeatureSpec',
    ['feature_type', 'feature_names', 'feature_ids', 'feature_is_request_only']
)

Definition at line 55 of file schema.py.