package tensorflow.data.model

Get desktop application:
View/edit binary Protocol Buffers messages

Algorithm used for model autotuning optimization.

Used in: AutotuneOptions, ModelProto.OptimizationParams

• DEFAULT = 0

• HILL_CLIMB = 1

• GRADIENT_DESCENT = 2

• MAX_PARALLELISM = 3

• STAGE_BASED = 4

Protocol buffer representing the data used by the autotuning modeling framework.

• map<int64, ModelProto.Node> nodes = 1

  Map of node IDs to nodes of this model.

• int64 output = 2

  ID of the output node of this model.

• int64 id_counter = 3

  Counter for node IDs of this model.

• optional ModelProto.OptimizationParams optimization_params = 5

General representation of a node in the model.

Used in: ModelProto

• int64 id = 1

  Unique node ID.

• string name = 2

  Human-readable name of the node.

• bool autotune = 3

  An indication whether autotuning is enabled for this node.

• int64 buffered_bytes = 4

  The number of bytes stored in this node's buffer.

• int64 buffered_elements = 5

  The number of elements stored in this node's buffer.

• int64 bytes_consumed = 6

  The number of bytes consumed by the node.

• int64 bytes_produced = 7

  The number of bytes produced by the node.

• int64 num_elements = 8

  The number of elements produced by the node.

• int64 processing_time = 9

  The aggregate processing time spent in this node in nanoseconds.

• bool record_metrics = 10

  An indication whether this node records metrics about produced and consumed elements.

• repeated Node.Parameter parameters = 11

  Parameters of this node.

• double input_processing_time_sum = 12

  Statistic of inputs processing time history.

• int64 input_processing_time_count = 13

• repeated int64 inputs = 14

  IDs of inputs of this node.

• NodeClass node_class = 15

  Class of this node.

• double ratio = 16

  Ratio of input to output elements. This is only used by KNOWN_RATIO and ASYNC_KNOWN_RATIO nodes.

• double memory_ratio = 17

  Ratio identifies how many parallelism calls are introduced by one buffered element. This is only used by ASYNC_KNOWN_RATIO nodes.

Represents a node parameter.

Used in: Node

• string name = 1

  Human-readable name of the parameter.

• double value = 2

  Identifies the model value of the parameter. This can be different from the actual value (e.g. during optimization search).

• double state_value = 3

  The actual value of the parameter.

• double min = 4

  Minimum value of the parameter.

• double max = 5

  Maximum value of the parameter.

• bool tunable = 6

  Identifies whether the parameter should participate in autotuning.

Contains parameters of the model autotuning optimization.

Used in: ModelProto

• AutotuneAlgorithm algorithm = 1

  Algorithm used for autotuning optimization.

• int64 cpu_budget = 2

  Number of available logical threads.

• int64 ram_budget = 3

  Amount of available memory in bytes.

• double model_input_time = 4

  Time between two consecutive `GetNext` calls to the iterator represented by the output node.

Class of a node in the performance model.

Used in: ModelProto.Node

• UNKNOWN = 0

• INTERLEAVE_MANY = 1

• ASYNC_INTERLEAVE_MANY = 2

• KNOWN_RATIO = 3

• ASYNC_KNOWN_RATIO = 4

• UNKNOWN_RATIO = 5

• ASYNC_UNKNOWN_RATIO = 6