STT-tensorflow/tensorflow/core/protobuf/struct.proto

syntax = "proto3";

package tensorflow;

import "tensorflow/core/framework/tensor.proto";
import "tensorflow/core/framework/tensor_shape.proto";
import "tensorflow/core/framework/types.proto";

option go_package = "github.com/tensorflow/tensorflow/tensorflow/go/core/core_protos_go_proto";

// `StructuredValue` represents a dynamically typed value representing various
// data structures that are inspired by Python data structures typically used in
// TensorFlow functions as inputs and outputs.
//
// For example when saving a Layer there may be a `training` argument. If the
// user passes a boolean True/False, that switches between two concrete
// TensorFlow functions. In order to switch between them in the same way after
// loading the SavedModel, we need to represent "True" and "False".
//
// A more advanced example might be a function which takes a list of
// dictionaries mapping from strings to Tensors. In order to map from
// user-specified arguments `[{"a": tf.constant(1.)}, {"q": tf.constant(3.)}]`
// after load to the right saved TensorFlow function, we need to represent the
// nested structure and the strings, recording that we have a trace for anything
// matching `[{"a": tf.TensorSpec(None, tf.float32)}, {"q": tf.TensorSpec([],
// tf.float64)}]` as an example.
//
// Likewise functions may return nested structures of Tensors, for example
// returning a dictionary mapping from strings to Tensors. In order for the
// loaded function to return the same structure we need to serialize it.
//
// This is an ergonomic aid for working with loaded SavedModels, not a promise
// to serialize all possible function signatures. For example we do not expect
// to pickle generic Python objects, and ideally we'd stay language-agnostic.
message StructuredValue {
  // The kind of value.
  oneof kind {
    // Represents None.
    NoneValue none_value = 1;

    // Represents a double-precision floating-point value (a Python `float`).
    double float64_value = 11;
    // Represents a signed integer value, limited to 64 bits.
    // Larger values from Python's arbitrary-precision integers are unsupported.
    sint64 int64_value = 12;
    // Represents a string of Unicode characters stored in a Python `str`.
    // In Python 3, this is exactly what type `str` is.
    // In Python 2, this is the UTF-8 encoding of the characters.
    // For strings with ASCII characters only (as often used in TensorFlow code)
    // there is effectively no difference between the language versions.
    // The obsolescent `unicode` type of Python 2 is not supported here.
    string string_value = 13;
    // Represents a boolean value.
    bool bool_value = 14;

    // Represents a TensorShape.
    tensorflow.TensorShapeProto tensor_shape_value = 31;
    // Represents an enum value for dtype.
    tensorflow.DataType tensor_dtype_value = 32;
    // Represents a value for tf.TensorSpec.
    TensorSpecProto tensor_spec_value = 33;
    // Represents a value for tf.TypeSpec.
    TypeSpecProto type_spec_value = 34;
    // Represents a value for tf.BoundedTensorSpec.
    BoundedTensorSpecProto bounded_tensor_spec_value = 35;

    // Represents a list of `Value`.
    ListValue list_value = 51;
    // Represents a tuple of `Value`.
    TupleValue tuple_value = 52;
    // Represents a dict `Value`.
    DictValue dict_value = 53;
    // Represents Python's namedtuple.
    NamedTupleValue named_tuple_value = 54;
  }
}

// Represents None.
message NoneValue {}

// Represents a Python list.
message ListValue {
  repeated StructuredValue values = 1;
}

// Represents a Python tuple.
message TupleValue {
  repeated StructuredValue values = 1;
}

// Represents a Python dict keyed by `str`.
// The comment on Unicode from Value.string_value applies analogously.
message DictValue {
  map<string, StructuredValue> fields = 1;
}

// Represents a (key, value) pair.
message PairValue {
  string key = 1;
  StructuredValue value = 2;
}

// Represents Python's namedtuple.
message NamedTupleValue {
  string name = 1;
  repeated PairValue values = 2;
}

// A protobuf to represent tf.TensorSpec.
message TensorSpecProto {
  string name = 1;
  tensorflow.TensorShapeProto shape = 2;
  tensorflow.DataType dtype = 3;
}

// A protobuf to represent tf.BoundedTensorSpec.
message BoundedTensorSpecProto {
  string name = 1;
  tensorflow.TensorShapeProto shape = 2;
  tensorflow.DataType dtype = 3;
  tensorflow.TensorProto minimum = 4;
  tensorflow.TensorProto maximum = 5;
}

// Represents a tf.TypeSpec
message TypeSpecProto {
  enum TypeSpecClass {
    UNKNOWN = 0;
    SPARSE_TENSOR_SPEC = 1;   // tf.SparseTensorSpec
    INDEXED_SLICES_SPEC = 2;  // tf.IndexedSlicesSpec
    RAGGED_TENSOR_SPEC = 3;   // tf.RaggedTensorSpec
    TENSOR_ARRAY_SPEC = 4;    // tf.TensorArraySpec
    DATA_DATASET_SPEC = 5;    // tf.data.DatasetSpec
    DATA_ITERATOR_SPEC = 6;   // IteratorSpec from data/ops/iterator_ops.py
    OPTIONAL_SPEC = 7;        // tf.OptionalSpec
    PER_REPLICA_SPEC = 8;     // PerReplicaSpec from distribute/values.py
    VARIABLE_SPEC = 9;        // tf.VariableSpec
    ROW_PARTITION_SPEC = 10;  // RowPartitionSpec from ragged/row_partition.py
  }
  TypeSpecClass type_spec_class = 1;

  // The value returned by TypeSpec._serialize().
  StructuredValue type_state = 2;

  // This is currently redundant with the type_spec_class enum, and is only
  // used for error reporting.  In particular, if you use an older binary to
  // load a newer model, and the model uses a TypeSpecClass that the older
  // binary doesn't support, then this lets us display a useful error message.
  string type_spec_class_name = 3;
}