package serialization

Get desktop application:
View/edit binary Protocol Buffers messages

Used in: BigDLModule, NameAttrList

• DataType dataType = 1

• string subType = 2

  specific for custom data

• oneof value

  • int32 int32Value = 3

    int32 value

  • int64 int64Value = 4

    int64 value

  • float floatValue = 5

    float value

  • double doubleValue = 6

    float value

  • string stringValue = 7

    string value

  • bool boolValue = 8

    bool value

  • Regularizer regularizerValue = 9

    Regularizer

  • BigDLTensor tensorValue = 10

    tensor value

  • VarFormat variableFormatValue = 11

    Variable format

  • InitMethod initMethodValue = 12

    init method

  • BigDLModule bigDLModuleValue = 13

    big DL module

  • NameAttrList nameAttrListValue = 14

    name attribute list

  • AttrValue.ArrayValue arrayValue = 15

    array value of any type

  • InputDataFormat dataFormatValue = 16

    data format

  • google.protobuf.Any customValue = 17

    custom value

Used in: AttrValue

• int32 size = 1

• DataType datatype = 2

• repeated int32 i32 = 3

  "Array(int32)"

• repeated int64 i64 = 4

  "Array(int64)"

• repeated float flt = 5

  "Array(float)"

• repeated double dbl = 6

  "Array(double)"

• repeated string str = 7

  "Array(string)"

• repeated bool boolean = 8

  "Array(bool)"

• repeated Regularizer Regularizer = 9

  "Array(Regularizer)"

• repeated BigDLTensor tensor = 10

  "Array(BigDLTensor)"

• repeated VarFormat variableFormat = 11

  "Array(VariableFormat)"

• repeated InitMethod initMethod = 12

  "Array(BigDLTensor)"

• repeated BigDLModule bigDLModule = 13

  "Array(BigDLModel)"

• repeated NameAttrList nameAttrList = 14

• repeated InputDataFormat dataFormat = 15

  "Array(DataFormat)"

• repeated google.protobuf.Any custom = 16

  "Array(Any)"

Used in: AttrValue, AttrValue.ArrayValue

• string name = 1

  module name

• repeated BigDLModule subModules = 2

  sub modules

• optional BigDLTensor weight = 3

  weight for each layer, serialized data are stored as either float or double

• optional BigDLTensor bias = 4

  bias for each layer

• repeated string preModules = 5

  pre modules of the same hierarchy

• repeated string nextModules = 6

  next modules of the same hierachy

• string moduleType = 7

  module type to identify a module like linear, graph, etc

• map<string, AttrValue> attr = 8

  module attributes

• string version = 9

  version of this module

• bool train = 10

  is module status in train

• string namePostfix = 11

  name post fix

• int32 id = 12

  unique ID of this module , used for shared modules

Used in: AttrValue, AttrValue.ArrayValue, BigDLModule

• DataType datatype = 1

• repeated int32 size = 2

  size of tensor

• repeated int32 stride = 3

  stride of tensor

• int32 offset = 4

  offset

• int32 dimension = 5

  dimension

• int32 nElements = 6

  total number of elements

• bool isScalar = 7

  is scalar tensor

• optional TensorStorage storage = 8

  reference to storage

• int32 id = 9

  tensor ID, used for tensor sharing

• TensorType tensorType = 10

Used in: AttrValue, AttrValue.ArrayValue, BigDLTensor, TensorStorage

• INT32 = 0

• INT64 = 1

• FLOAT = 2

• DOUBLE = 3

• STRING = 4

• BOOL = 5

• CHAR = 6

• SHORT = 7

• BYTES = 8

• REGULARIZER = 9

• TENSOR = 10

• VARIABLE_FORMAT = 11

• INITMETHOD = 12

• MODULE = 13

• NAME_ATTR_LIST = 14

• ARRAY_VALUE = 15

• DATA_FORMAT = 16

• CUSTOM = 17

Used in: AttrValue, AttrValue.ArrayValue

• InitMethodType methodType = 1

• repeated double data = 2

Used in: InitMethod

• EMPTY_INITIALIZATION = 0

• RANDOM_UNIFORM = 1

• RANDOM_UNIFORM_PARAM = 2

• RANDOM_NORMAL = 3

• ZEROS = 4

• ONES = 5

• CONST = 6

• XAVIER = 7

• BILINEARFILLER = 8

Used in: AttrValue, AttrValue.ArrayValue

• NCHW = 0

• NHWC = 1

Used in: AttrValue, AttrValue.ArrayValue

• string name = 1

• map<string, AttrValue> attr = 2

Used in: AttrValue, AttrValue.ArrayValue

• RegularizerType regularizerType = 1

• repeated double regularData = 2

Used in: Regularizer

• L1L2Regularizer = 0

• L1Regularizer = 1

• L2Regularizer = 2

Used in: BigDLTensor

• DataType datatype = 1

• repeated float float_data = 2

  float tensor element

• repeated double double_data = 3

  double tensor element

• repeated bool bool_data = 4

  boolean tensor element

• repeated string string_data = 5

  string tensor element

• repeated int32 int_data = 6

  int tensor element

• repeated int64 long_data = 7

  long tensor element

• repeated bytes bytes_data = 8

  byte tensor element

• int32 id = 9

  storage Id, used for storage sharing

Used in: BigDLTensor

• DENSE = 0

• QUANT = 1

Used in: AttrValue, AttrValue.ArrayValue

• EMPTY_FORMAT = 0

• DEFAULT = 1

• ONE_D = 2

• IN_OUT = 3

• OUT_IN = 4

• IN_OUT_KW_KH = 5

• OUT_IN_KW_KH = 6

• GP_OUT_IN_KW_KH = 7

• GP_IN_OUT_KW_KH = 8

• OUT_IN_KT_KH_KW = 9