tensorflow

package
v2.14.1 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Dec 8, 2023 License: BSD-3-Clause Imports: 7 Imported by: 0

Documentation

Index

Constants

This section is empty.

Variables

View Source 
var File_tensorflow_serving_servables_tensorflow_saved_model_bundle_source_adapter_proto protoreflect.FileDescriptor
View Source 
var File_tensorflow_serving_servables_tensorflow_session_bundle_config_proto protoreflect.FileDescriptor
View Source 
var File_tensorflow_serving_servables_tensorflow_thread_pool_factory_config_proto protoreflect.FileDescriptor

Functions

This section is empty.

Types

type BatchingParameters 

type BatchingParameters struct {

	// The maximum size of each input batch.
	//
	// IMPORTANT: As discussed above, use 'max_batch_size * 2' client threads to
	// achieve high throughput with batching.
	MaxBatchSize *wrapperspb.Int64Value `protobuf:"bytes,1,opt,name=max_batch_size,json=maxBatchSize,proto3" json:"max_batch_size,omitempty"`
	// If a task has been enqueued for this amount of time (in microseconds), and
	// a thread is available, the scheduler will immediately form a batch from
	// enqueued tasks and assign the batch to the thread for processing, even if
	// the batch's size is below 'max_batch_size'.
	BatchTimeoutMicros *wrapperspb.Int64Value `protobuf:"bytes,2,opt,name=batch_timeout_micros,json=batchTimeoutMicros,proto3" json:"batch_timeout_micros,omitempty"`
	// The maximum length of the queue, in terms of the number of batches. (A
	// batch that has been scheduled on a thread is considered to have been
	// removed from the queue.)
	MaxEnqueuedBatches *wrapperspb.Int64Value `protobuf:"bytes,3,opt,name=max_enqueued_batches,json=maxEnqueuedBatches,proto3" json:"max_enqueued_batches,omitempty"`
	// The number of threads to use to process batches.
	// Must be >= 1, and should be tuned carefully.
	NumBatchThreads *wrapperspb.Int64Value `protobuf:"bytes,4,opt,name=num_batch_threads,json=numBatchThreads,proto3" json:"num_batch_threads,omitempty"`
	// The name to use for the pool of batch threads.
	ThreadPoolName *wrapperspb.StringValue `protobuf:"bytes,5,opt,name=thread_pool_name,json=threadPoolName,proto3" json:"thread_pool_name,omitempty"`
	// If true, queue implementation would split one input batch task into
	// subtasks (as specified by `split_input_task_func` below) and fit subtasks
	// into different batches.
	EnableLargeBatchSplitting *wrapperspb.BoolValue `` /* 140-byte string literal not displayed */
	// The maximum size of each enqueued batch to be processed (i.e., in
	// `batches_`). Relevant iff enable_large_batch_splitting is true. And when
	// relevant, 'max_batch_size' should be greater or equal than
	// `max_execution_batch_size`
	//
	// The scheduler may form batches of any size between 1 and this number
	// (inclusive).
	MaxExecutionBatchSize *wrapperspb.Int64Value `` /* 128-byte string literal not displayed */
	// The allowed batch sizes. (Ignored if left empty.)
	// Requirements:
	//   - The entries must be in increasing order.
	//   - The final entry must equal 'max_batch_size'.
	AllowedBatchSizes []int64 `protobuf:"varint,6,rep,packed,name=allowed_batch_sizes,json=allowedBatchSizes,proto3" json:"allowed_batch_sizes,omitempty"`
	// Whether to pad variable-length inputs when a batch is formed.
	PadVariableLengthInputs bool `` /* 135-byte string literal not displayed */
	// contains filtered or unexported fields
}

Batching parameters. Each individual parameter is optional. If omitted, the default value from the relevant batching config struct (SharedBatchScheduler ::Options or BatchSchedulerRetrier::Options) is used.

func (*BatchingParameters) Descriptor deprecated 

func (*BatchingParameters) Descriptor() ([]byte, []int)

Deprecated: Use BatchingParameters.ProtoReflect.Descriptor instead.

func (*BatchingParameters) GetAllowedBatchSizes 

func (x *BatchingParameters) GetAllowedBatchSizes() []int64

func (*BatchingParameters) GetBatchTimeoutMicros 

func (x *BatchingParameters) GetBatchTimeoutMicros() *wrapperspb.Int64Value

func (*BatchingParameters) GetEnableLargeBatchSplitting 

func (x *BatchingParameters) GetEnableLargeBatchSplitting() *wrapperspb.BoolValue

func (*BatchingParameters) GetMaxBatchSize 

func (x *BatchingParameters) GetMaxBatchSize() *wrapperspb.Int64Value

func (*BatchingParameters) GetMaxEnqueuedBatches 

func (x *BatchingParameters) GetMaxEnqueuedBatches() *wrapperspb.Int64Value

func (*BatchingParameters) GetMaxExecutionBatchSize 

func (x *BatchingParameters) GetMaxExecutionBatchSize() *wrapperspb.Int64Value

func (*BatchingParameters) GetNumBatchThreads 

func (x *BatchingParameters) GetNumBatchThreads() *wrapperspb.Int64Value

func (*BatchingParameters) GetPadVariableLengthInputs 

func (x *BatchingParameters) GetPadVariableLengthInputs() bool

func (*BatchingParameters) GetThreadPoolName 

func (x *BatchingParameters) GetThreadPoolName() *wrapperspb.StringValue

func (*BatchingParameters) ProtoMessage 

func (*BatchingParameters) ProtoMessage()

func (*BatchingParameters) ProtoReflect 

func (x *BatchingParameters) ProtoReflect() protoreflect.Message

func (*BatchingParameters) Reset 

func (x *BatchingParameters) Reset()

func (*BatchingParameters) String 

func (x *BatchingParameters) String() string

type ModelWarmupOptions 

type ModelWarmupOptions struct {

	// Number of times a request is iterated during warmup replay. By default 1.
	NumRequestIterations *wrapperspb.Int32Value `protobuf:"bytes,1,opt,name=num_request_iterations,json=numRequestIterations,proto3" json:"num_request_iterations,omitempty"`
	// The number of threads to parallel execute warm up queries. By default 1.
	NumModelWarmupThreads *wrapperspb.Int32Value `` /* 128-byte string literal not displayed */
	// Model name.
	ModelName string `protobuf:"bytes,3,opt,name=model_name,json=modelName,proto3" json:"model_name,omitempty"`
	// Model version.
	ModelVersion int64 `protobuf:"varint,4,opt,name=model_version,json=modelVersion,proto3" json:"model_version,omitempty"`
	// If true, warmup queries initiate parallel (dummy) warmup queries for each
	// `allowed_batch_sizes` of supported batch ops.
	// The extra queries' outputs are not returned.
	EnableAllBatchSizesWarmup bool `` /* 143-byte string literal not displayed */
	// contains filtered or unexported fields
}

Options related to model-warmup.

func (*ModelWarmupOptions) Descriptor deprecated 

func (*ModelWarmupOptions) Descriptor() ([]byte, []int)

Deprecated: Use ModelWarmupOptions.ProtoReflect.Descriptor instead.

func (*ModelWarmupOptions) GetEnableAllBatchSizesWarmup added in v2.14.0 

func (x *ModelWarmupOptions) GetEnableAllBatchSizesWarmup() bool

func (*ModelWarmupOptions) GetModelName added in v2.14.0 

func (x *ModelWarmupOptions) GetModelName() string

func (*ModelWarmupOptions) GetModelVersion added in v2.14.0 

func (x *ModelWarmupOptions) GetModelVersion() int64

func (*ModelWarmupOptions) GetNumModelWarmupThreads 

func (x *ModelWarmupOptions) GetNumModelWarmupThreads() *wrapperspb.Int32Value

func (*ModelWarmupOptions) GetNumRequestIterations 

func (x *ModelWarmupOptions) GetNumRequestIterations() *wrapperspb.Int32Value

func (*ModelWarmupOptions) ProtoMessage 

func (*ModelWarmupOptions) ProtoMessage()

func (*ModelWarmupOptions) ProtoReflect 

func (x *ModelWarmupOptions) ProtoReflect() protoreflect.Message

func (*ModelWarmupOptions) Reset 

func (x *ModelWarmupOptions) Reset()

func (*ModelWarmupOptions) String 

func (x *ModelWarmupOptions) String() string

type SavedModelBundleSourceAdapterConfig 

type SavedModelBundleSourceAdapterConfig struct {

	// A SessionBundleConfig.
	// FOR INTERNAL USE ONLY DURING TRANSITION TO SAVED_MODEL. WILL BE DEPRECATED.
	// TODO(b/32248363): Replace this field with the "real" field(s).
	LegacyConfig *SessionBundleConfig `protobuf:"bytes,1000,opt,name=legacy_config,json=legacyConfig,proto3" json:"legacy_config,omitempty"`
	// contains filtered or unexported fields
}

Config proto for SavedModelBundleSourceAdapter.

func (*SavedModelBundleSourceAdapterConfig) Descriptor deprecated 

func (*SavedModelBundleSourceAdapterConfig) Descriptor() ([]byte, []int)

Deprecated: Use SavedModelBundleSourceAdapterConfig.ProtoReflect.Descriptor instead.

func (*SavedModelBundleSourceAdapterConfig) GetLegacyConfig 

func (x *SavedModelBundleSourceAdapterConfig) GetLegacyConfig() *SessionBundleConfig

func (*SavedModelBundleSourceAdapterConfig) ProtoMessage 

func (*SavedModelBundleSourceAdapterConfig) ProtoMessage()

func (*SavedModelBundleSourceAdapterConfig) ProtoReflect 

func (x *SavedModelBundleSourceAdapterConfig) ProtoReflect() protoreflect.Message

func (*SavedModelBundleSourceAdapterConfig) Reset 

func (x *SavedModelBundleSourceAdapterConfig) Reset()

func (*SavedModelBundleSourceAdapterConfig) String 

func (x *SavedModelBundleSourceAdapterConfig) String() string

type SessionBundleConfig 

type SessionBundleConfig struct {

	// The TensorFlow runtime to connect to.
	// See full documentation in tensorflow/core/public/session_options.h.
	//
	// For single machine serving, we recommend using the empty string "", which
	// will configure the local TensorFlow runtime implementation. This provides
	// the best isolation currently available across multiple Session servables.
	SessionTarget string `protobuf:"bytes,1,opt,name=session_target,json=sessionTarget,proto3" json:"session_target,omitempty"`
	// TensorFlow Session configuration options.
	// See details at tensorflow/core/protobuf/config.proto.
	SessionConfig *protobuf.ConfigProto `protobuf:"bytes,2,opt,name=session_config,json=sessionConfig,proto3" json:"session_config,omitempty"`
	// If set, each emitted session is wrapped with a layer that schedules Run()
	// calls in batches. The batching layer is transparent to the client
	// (implements the tensorflow::Session API).
	//
	// IMPORTANT: With batching enabled, client threads will spend most of their
	// time blocked on Session::Run() calls, waiting for enough peer threads to
	// also call Session::Run() such that a large batch can be formed. For good
	// throughput, we recommend setting the number of client threads equal to
	// roughly twice the maximum batch size ('max_batch_size' below).
	//
	// The batching layer uses a SharedBatchScheduler to coordinate batching
	// across multiple session servables emitted by this source adapter. A
	// BatchSchedulerRetrier is added on top of each batching session.
	BatchingParameters *BatchingParameters `protobuf:"bytes,3,opt,name=batching_parameters,json=batchingParameters,proto3" json:"batching_parameters,omitempty"`
	// If set, session run calls use a separate threadpool for restore and init
	// ops as part of loading the session-bundle. The value of this field should
	// correspond to the index of the tensorflow::ThreadPoolOptionProto defined as
	// part of `session_config.session_inter_op_thread_pool`.
	SessionRunLoadThreadpoolIndex *wrapperspb.Int32Value `` /* 154-byte string literal not displayed */
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION.
	//
	// Transient memory used while loading a model, which is released once the
	// loading phase has completed. (This is on top of the memory used in steady-
	// state while the model is in memory after it has finished loading.)
	//
	// TODO(b/38376838): This is a temporary hack, and it applies to all models.
	// Remove it once resource estimates are moved inside SavedModel.
	ExperimentalTransientRamBytesDuringLoad uint64 `` /* 187-byte string literal not displayed */
	// Set of SavedModel tags identifying the specific meta graph def to be
	// loaded.
	SavedModelTags []string `protobuf:"bytes,6,rep,name=saved_model_tags,json=savedModelTags,proto3" json:"saved_model_tags,omitempty"`
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION.
	//
	// Input tensors to append to every Session::Run() call.
	ExperimentalFixedInputTensors []*protobuf.NamedTensorProto `` /* 154-byte string literal not displayed */
	// Enables model warmup.
	EnableModelWarmup  bool                `protobuf:"varint,779,opt,name=enable_model_warmup,json=enableModelWarmup,proto3" json:"enable_model_warmup,omitempty"`
	ModelWarmupOptions *ModelWarmupOptions `protobuf:"bytes,780,opt,name=model_warmup_options,json=modelWarmupOptions,proto3" json:"model_warmup_options,omitempty"`
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION.
	//
	// Enables passing in the SessionMetadata to the Session. The SessionMetadata
	// consists of information like the model name, version, which can then be
	// used by the TensorFlow runtime appropriately (for debugging, logging, etc).
	EnableSessionMetadata bool `` /* 129-byte string literal not displayed */
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION.
	//
	// Remove unused fields from metagraph proto message in SavedModelBundle.
	// This message is stored alongside the `Session` object. Removing unwanted
	// fields helps reduce memory footprint.
	RemoveUnusedFieldsFromBundleMetagraph bool `` /* 183-byte string literal not displayed */
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION.
	//
	// Prefer TensorFlow Lite model from `model.tflite` file in SavedModel
	// directory, instead of the TensorFlow model from `saved_model.pb` file.
	// If no TensorFlow Lite model found, fallback to TensorFlow model.
	PreferTfliteModel bool `protobuf:"varint,783,opt,name=prefer_tflite_model,json=preferTfliteModel,proto3" json:"prefer_tflite_model,omitempty"`
	// Tries to use infra validation result to estimate resource usage.
	ResourceEstimationUsesValidationResult bool `` /* 184-byte string literal not displayed */
	// Deprecated: Marked as deprecated in tensorflow_serving/servables/tensorflow/session_bundle_config.proto.
	NumTfliteInterpreters int32 `` /* 129-byte string literal not displayed */
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION.
	//
	// Number of TFLite interpreters in an interpreter pool of TfLiteSession.
	NumTfliteInterpretersPerPool int32 `` /* 154-byte string literal not displayed */
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION.
	//
	// Number of TFLite interpreter pools in a TfLiteSession.
	NumTflitePools int32 `protobuf:"varint,787,opt,name=num_tflite_pools,json=numTflitePools,proto3" json:"num_tflite_pools,omitempty"`
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION.
	//
	// Use SessionWrapperIgnoreThreadPoolOptions instead.
	WrapSessionWithNoThreadingParams bool `` /* 168-byte string literal not displayed */
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION.
	//
	// Enable per-model batching parameters (present in SavedModel). If this
	// option is enabled, model specific batching params (e.g. timeout, batch
	// sizes etc.) from `batching_parameters` field above are *ignored* and
	// instead the one in SavedModel directory are used. This field is only
	// used if batching is enabled (i.e. `batching_parameters` message above
	// is set).
	EnablePerModelBatchingParams bool `` /* 154-byte string literal not displayed */
	// EXPERIMENTAL. THIS FIELD MAY CHANGE OR GO AWAY. USE WITH CAUTION
	//
	// Enable overriding model configs via assets.extra/saved_model_config.pb.
	EnableSavedModelConfig bool `` /* 134-byte string literal not displayed */
	// contains filtered or unexported fields
}

Configuration parameters for a SessionBundle, with optional batching.

func (*SessionBundleConfig) Descriptor deprecated 

func (*SessionBundleConfig) Descriptor() ([]byte, []int)

Deprecated: Use SessionBundleConfig.ProtoReflect.Descriptor instead.

func (*SessionBundleConfig) GetBatchingParameters 

func (x *SessionBundleConfig) GetBatchingParameters() *BatchingParameters

func (*SessionBundleConfig) GetEnableModelWarmup 

func (x *SessionBundleConfig) GetEnableModelWarmup() bool

func (*SessionBundleConfig) GetEnablePerModelBatchingParams added in v2.11.0 

func (x *SessionBundleConfig) GetEnablePerModelBatchingParams() bool

func (*SessionBundleConfig) GetEnableSavedModelConfig added in v2.14.0 

func (x *SessionBundleConfig) GetEnableSavedModelConfig() bool

func (*SessionBundleConfig) GetEnableSessionMetadata 

func (x *SessionBundleConfig) GetEnableSessionMetadata() bool

func (*SessionBundleConfig) GetExperimentalFixedInputTensors 

func (x *SessionBundleConfig) GetExperimentalFixedInputTensors() []*protobuf.NamedTensorProto

func (*SessionBundleConfig) GetExperimentalTransientRamBytesDuringLoad 

func (x *SessionBundleConfig) GetExperimentalTransientRamBytesDuringLoad() uint64

func (*SessionBundleConfig) GetModelWarmupOptions 

func (x *SessionBundleConfig) GetModelWarmupOptions() *ModelWarmupOptions

func (*SessionBundleConfig) GetNumTfliteInterpreters deprecated 

func (x *SessionBundleConfig) GetNumTfliteInterpreters() int32

Deprecated: Marked as deprecated in tensorflow_serving/servables/tensorflow/session_bundle_config.proto.

func (*SessionBundleConfig) GetNumTfliteInterpretersPerPool 

func (x *SessionBundleConfig) GetNumTfliteInterpretersPerPool() int32

func (*SessionBundleConfig) GetNumTflitePools 

func (x *SessionBundleConfig) GetNumTflitePools() int32

func (*SessionBundleConfig) GetPreferTfliteModel 

func (x *SessionBundleConfig) GetPreferTfliteModel() bool

func (*SessionBundleConfig) GetRemoveUnusedFieldsFromBundleMetagraph 

func (x *SessionBundleConfig) GetRemoveUnusedFieldsFromBundleMetagraph() bool

func (*SessionBundleConfig) GetResourceEstimationUsesValidationResult 

func (x *SessionBundleConfig) GetResourceEstimationUsesValidationResult() bool

func (*SessionBundleConfig) GetSavedModelTags 

func (x *SessionBundleConfig) GetSavedModelTags() []string

func (*SessionBundleConfig) GetSessionConfig 

func (x *SessionBundleConfig) GetSessionConfig() *protobuf.ConfigProto

func (*SessionBundleConfig) GetSessionRunLoadThreadpoolIndex 

func (x *SessionBundleConfig) GetSessionRunLoadThreadpoolIndex() *wrapperspb.Int32Value

func (*SessionBundleConfig) GetSessionTarget 

func (x *SessionBundleConfig) GetSessionTarget() string

func (*SessionBundleConfig) GetWrapSessionWithNoThreadingParams 

func (x *SessionBundleConfig) GetWrapSessionWithNoThreadingParams() bool

func (*SessionBundleConfig) ProtoMessage 

func (*SessionBundleConfig) ProtoMessage()

func (*SessionBundleConfig) ProtoReflect 

func (x *SessionBundleConfig) ProtoReflect() protoreflect.Message

func (*SessionBundleConfig) Reset 

func (x *SessionBundleConfig) Reset()

func (*SessionBundleConfig) String 

func (x *SessionBundleConfig) String() string

type ThreadPoolFactoryConfig 

type ThreadPoolFactoryConfig struct {

	// The config proto for a ThreadPoolFactory in the ThreadPoolFactory registry.
	ThreadPoolFactoryConfig *anypb.Any `` /* 134-byte string literal not displayed */
	// contains filtered or unexported fields
}

Configuration for a thread pool factory.

func (*ThreadPoolFactoryConfig) Descriptor deprecated 

func (*ThreadPoolFactoryConfig) Descriptor() ([]byte, []int)

Deprecated: Use ThreadPoolFactoryConfig.ProtoReflect.Descriptor instead.

func (*ThreadPoolFactoryConfig) GetThreadPoolFactoryConfig 

func (x *ThreadPoolFactoryConfig) GetThreadPoolFactoryConfig() *anypb.Any

func (*ThreadPoolFactoryConfig) ProtoMessage 

func (*ThreadPoolFactoryConfig) ProtoMessage()

func (*ThreadPoolFactoryConfig) ProtoReflect 

func (x *ThreadPoolFactoryConfig) ProtoReflect() protoreflect.Message

func (*ThreadPoolFactoryConfig) Reset 

func (x *ThreadPoolFactoryConfig) Reset()

func (*ThreadPoolFactoryConfig) String 

func (x *ThreadPoolFactoryConfig) String() string

Source Files

View all Source files

Directories

Path Synopsis

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.