MLOps gRPC Gateway

sasContainer (optional)

String The name of the container from which to download blobs.

storageConnectionString (optional)

String A storage account connection string.

`DeploymentStatusDeploymentState` - Possible progression of states: Up

PREPARING ::> LAUNCHING | FAILED LAUNCHING ::> HEALTHY | FAILED FAILED ::> PREPARING HEALTHY ::> UNHEALTHY | TERMINATING UNHEALTHY ::> HEALTHY | TERMINATING TERMINATING ::> TERMINATED TERMINATED ::> PREPARING (only when workspaces are undeleted/unarchived)

DEPLOYMENT_STATE_UNSPECIFIED: Unknown status
PREPARING: Preparing the Deployment for launch E.g., building assets.
LAUNCHING: Deployment is launching to an environment. E.g., waiting on target environment to spin up.
FAILED: Deployment failed during preparation or launch.
HEALTHY: Deployment is alive and healthy.
UNHEALTHY: Health issues detected with the launched Deployment.
TERMINATING: Deployment is terminating. I.e., target environment resources are being brought down.
PENDING: Deployment created and awaiting processing.
TERMINATED: Deployment is terminated.
STOPPED: Stopped (when deployments is scaled down)

`ExperimentExperimentStatus` - Up

`JobState` - Up

State represents the current state of the job.

STATE_UNSPECIFIED: State is unspecified or unknown.
CREATED: Job is created and waiting for submission
SUBMITTED: Job is submitted and waiting for starting up
RUNNING: Job is being executed.
PENDING: Job is waiting for scorer to be ready. Possible changes are from Submitted -> Pending or Running -> Pending.
FINISHED: Job has completed successfully.
FAILED: Job has completed with an error.
CANCELLED: Job has been cancelled.
SUBMISSION_FAILED: Job submission failed
CANCELLING: Job cancellation has been started.
TIMEOUT: Job has been time out

`MetricReferenceDatasetReference` - Up

TRAINING_DATASET: References a metric evaluated using the experiments training dataset. This can be different for each experiment.
TESTING_DATASET: References a metric evaluated using the experiments testing dataset. This can be different for each experiment.
VALIDATION_DATASET: References a metric evaluated using the experiments validation dataset. This can be different for each experiment.
CUSTOM_DATASET: References a metric evaluated using a particular dataset same for all the experiments. If chosen, the 'dataset' field must be set and contain id of that dataset.

`ModelRequestParametersShapleyType` - Enum defining the type of Shapley value calculation to be performed during scoring Up

SHAPLEY_TYPE_UNSPECIFIED: Default value. Server will choose.
NONE: No Shapley value calculation will be performed.
ORIGINAL: Calculate Shapley values using the original/raw feature values.
TRANSFORMED: Calculate Shapley values using transformed/preprocessed feature values.

`Stream_result_of_bytestreamReadResponse` - Stream result of bytestreamReadResponse Up

result (optional)

bytestreamReadResponse

error (optional)

rpcStatus

`Stream_result_of_v1JobOutputResponse` - Stream result of v1JobOutputResponse Up

result (optional)

v1JobOutputResponse

error (optional)

rpcStatus

`The_deployment_to_update_` - The deployment to update. Up

displayName (optional)

String Optional. Human-readable name.

description (optional)

String Optional. Description of the deployment.

etag (optional)

String The etag of the deployment.

createTime (optional)

Date Creation timestamp. format: date-time

updateTime (optional)

Date Update timestamp. format: date-time

creator (optional)

String Output only. Name of an entity that created the Deployment.

single (optional)

shadow (optional)

split (optional)

security (optional)

cors (optional)

v2Cors

monitoringOptions (optional)

v2MonitoringOptions

`The_endpoint_to_update__where_the_endpoint_snamefield_is_used_to_identify_the_one_to_update_` - The endpoint to update, where the endpoint's `name` field is used to identify the one to update. Up

displayName (optional)

String The human-readable name of this endpoint.

description (optional)

String The description of this endpoint.

path

String Required. The URI path for this endpoint. This must be in path-noscheme form without the leading, ending and multiple consecutive "/" (see https://datatracker.ietf.org/doc/html/rfc3986#appendix-A).

target (optional)

String If left out or empty, the endpoint is considered detached.

createTime (optional)

Date Output only. The time the endpoint was created. format: date-time

updateTime (optional)

Date Output only. The time the endpoint was last updated. format: date-time

`bytestreamQueryWriteStatusRequest` - Up

Request object for ByteStream.QueryWriteStatus.

resourceName (optional)

String The name of the resource whose write status is being requested.

`bytestreamQueryWriteStatusResponse` - Up

Response object for ByteStream.QueryWriteStatus.

committedSize (optional)

String The number of bytes that have been processed for the given resource. format: int64

complete (optional)

Boolean complete is true only if the client has sent a WriteRequest with finish_write set to true, and the server has processed that request.

`bytestreamReadRequest` - Up

Request object for ByteStream.Read.

resourceName (optional)

String The name of the resource to read.

readOffset (optional)

The offset for the first byte to return in the read, relative to the start of the resource.

A read_offset that is negative or greater than the size of the resource will cause an OUT_OF_RANGE error.

format: int64

readLimit (optional)

The maximum number of data bytes the server is allowed to return in the sum of all ReadResponse messages. A read_limit of zero indicates that there is no limit, and a negative read_limit will cause an error.

If the stream returns fewer bytes than allowed by the read_limit and no error occurred, the stream includes all data from the read_offset to the end of the resource.

format: int64

`bytestreamReadResponse` - Up

Response object for ByteStream.Read.

data (optional)

byte[] A portion of the data for the resource. The service may leave data empty for any given ReadResponse. This enables the service to inform the client that the request is still live while it is running an operation to generate more data. format: byte

`bytestreamWriteRequest` - Up

Request object for ByteStream.Write.

resourceName (optional)

String The name of the resource to write. This must be set on the first WriteRequest of each Write() action. If it is set on subsequent calls, it must match the value of the first request.

writeOffset (optional)

The offset from the beginning of the resource at which the data should be written. It is required on all WriteRequests.

In the first WriteRequest of a Write() action, it indicates the initial offset for the Write() call. The value must be equal to the committed_size that a call to QueryWriteStatus() would return.

On subsequent calls, this value must be set and must be equal to the sum of the first write_offset and the sizes of all data bundles sent previously on this stream.

An incorrect value will cause an error.

format: int64

finishWrite (optional)

Boolean If true, this indicates that the write is complete. Sending any WriteRequests subsequent to one in which finish_write is true will cause an error.

data (optional)

byte[] A portion of the data for the resource. The client may leave data empty for any given WriteRequest. This enables the client to inform the service that the request is still live while it is running an operation to generate more data. format: byte

`bytestreamWriteResponse` - Up

Response object for ByteStream.Write.

committedSize (optional)

String The number of bytes that have been processed for the given resource. format: int64

`ingestCreateModelIngestionRequest` - Up

artifactId (optional)

map[String, ingestMetadataValue]

`ingestCreateModelIngestionResponse` - Up

ingestion (optional)

ingestModelIngestion

`ingestMetadata` - Up

Container for metadata.

values (optional)

`ingestMetadataValue` - Up

Represents a generic holder for different value types for metadata.

doubleValue (optional)

Double format: double

stringValue (optional)

int64Value (optional)

String format: int64

boolValue (optional)

jsonValue (optional)

timestampValue (optional)

Date format: date-time

durationValue (optional)

`ingestModelIngestion` - Up

artifactId (optional)

String ID of the artifact ingested.

modelParameters (optional)

ingestModelParameters

modelMetadata (optional)

ingestMetadata

artifactType (optional)

String Artifact type.

`ingestModelParameters` - Up

Main parameters of the experiment.

targetColumn (optional)

String The column to predict.

`inline_object` - Up

The request message to get suggested resource for given composition.

runtime

workers (optional)

Integer Optional additional field The number of runtime worker processes (degree of parallelization), only applicable to python based scorers. If omit the value then it will fallback to 1. format: int32

`inline_object_1` - Up

The request message for generate logs.

sinceTime (optional)

Date (-- api-linter: core::0140::prepositions=disabled aip.dev/not-precedent: We need to do this because this name is the same in K8 logs API. --) Timestamp indicating time from logs should be retrieved. format: date-time

`protobufAny` - Up

Attype (optional)

`rpcStatus` - Up

code (optional)

Integer format: int32

message (optional)

details (optional)

array[protobufAny]

`storageArtifact` - Up

Represents a large immutable blob of data.

id (optional)

String Unique ID of the artifact.

entityId (optional)

String ID of the parent entity, e.g., an Experiment. Permissions of the parent entity are used to authorize users accessing the artifact.

type (optional)

String Defines type of the artifact in relation to the parent entity. Each entity type defines certain well-known types with universal meaning, e.g., "dataset" for the Dataset entity containing the actual tabular data. In addition, ML backends can define and store additional artifacts for each entity.

state (optional)

ArtifactArtifactState

size (optional)

String Size of the artifact data at rest in bytes. I.e., if the data is compressed, the size represents the size after compression. format: int64

md5Digest (optional)

byte[]

Fingerprint of the artifact data. This can be used, e.g., to see if an artifact needs re-uploading (as a new artifact due to immutability).

TODO(osery): Consider other options here. We don't need to be too flexible, but shouldn't be something too slow.

format: byte

createdTime (optional)

Date Creation timestamp. format: date-time

uploadedTime (optional)

Date End of upload timestamp. format: date-time

mimeType (optional)

String MIME type of the artifact data.

lastModifiedTime (optional)

Date Timestamp of last modification. format: date-time

entityType (optional)

storageEntityType

`storageAuditEvent` - Up

Represents a single event in the history of an entity.

id (optional)

String Unique ID of this event.

actorId (optional)

String ID of the user that has caused the event (e.g., by performing an operation or an action).

userId (optional)

String ID of the user this event affects (see the actor field for the user whose action has caused the event).

groupId (optional)

String ID of the user group this event is related to, if any.

projectId (optional)

String ID of the project this event is related to, if any.

experimentId (optional)

String ID of the experiment this event is related to, if any.

datasetId (optional)

String ID of the dataset this event is related to, if any.

tagId (optional)

String ID of the tag this event is related to, if any.

commentId (optional)

String ID of the comment this event is related to, if any.

deploymentId (optional)

String ID of the deployment this event is related to, if any.

deploymentEnvironmentId (optional)

String ID of the deployment environment this event is related to, if any.

restrictionRoleId (optional)

String For some SHARING events, this field captures the role restriction, if set.

message (optional)

String Textual message describing the event in a human readable form.

timestamp (optional)

Date Event timestamp. format: date-time

operation (optional)

storageOperation

registeredModelId (optional)

String ID of the registered model, if any.

registeredModelVersionId (optional)

array[storageDeleteDatasetRequest]

`storageBatchDeleteDatasetRequest` - Up

datasetRequest (optional)

`storageBatchDeleteDatasetResponse` - Up

datasetDeleteResponse (optional)

array[storageDeleteDatasetResponse]

`storageBatchDeleteExperimentRequest` - Up

experimentRequest (optional)

array[storageDeleteExperimentRequest]

`storageBatchDeleteExperimentResponse` - Up

experimentDeleteResponse (optional)

array[storageDeleteExperimentResponse]

`storageBatchDeleteModelVersionRequest` - Up

ids (optional)

array[storageDeleteModelVersionResponse]

`storageBatchDeleteModelVersionResponse` - Up

deleteModelVersionResponse (optional)

`storageBatchDeleteProjectRequest` - Up

projectIds (optional)

array[String] List of Project IDs to be deleted.

`storageBatchDeleteProjectResponse` - Up

projectsDeleted (optional)

array[storageDeleteProjectResponse]

`storageBatchDeleteRegisteredModelRequest` - Up

modelIds (optional)

array[String] IDs of the registered models to be deleted.

`storageBatchDeleteRegisteredModelResponse` - Up

deleteRegisteredModelResponse (optional)

array[storageDeleteRegisteredModelResponse]

`storageBatchRecoverDeletedExperimentRequest` - Up

recoverExperimentRequests (optional)

array[storageRecoverDeletedExperimentRequest]

`storageBatchRecoverDeletedExperimentResponse` - Up

recoverExperimentResponses (optional)

array[storageRecoverDeletedExperimentResponse]

`storageClause` - Up

Conjunction of multiple constraints. There is an implicit conjunction, AND, between all the constraints inside.

propertyConstraint (optional)

array[storagePropertyConstraint] Constraints on a values of an entity properties.

tagConstraint (optional)

array[storageTagConstraint] Constraints on a entity tags.

`storageColumn` - Up

name (optional)

array[storageHistogramEntry]

type (optional)

storageDataType

stats (optional)

storageColumnStats

sampleData (optional)

array[storageDataValue]

`storageColumnStats` - Up

count (optional)

String format: int64

unique (optional)

String format: int64

freq (optional)

String format: int64

numClasses (optional)

String format: int64

histogram (optional)

numeric (optional)

storageNumericColumnStats

nonNumeric (optional)

storageNonNumericColumnStats

time (optional)

storageTimeColumnStats

`storageComment` - Up

Represents a single comment.

id (optional)

String Unique ID of this comment.

authorId (optional)

String ID of the user that has created the comment.

projectId (optional)

String ID of the project this comment is attached to, if any.

experimentId (optional)

String ID of the experiment this comment is attached to, if any.

datasetId (optional)

String ID of the dataset this comment is attached to, if any.

message (optional)

String Textual payload of the comment.

createdTime (optional)

Date Creation timestamp. format: date-time

`storageCountDeploymentsRequest` - Up

projectId (optional)

filter (optional)

`storageCountDeploymentsResponse` - Up

count (optional)

String format: int64

`storageCountExperimentsRequest` - Up

filter (optional)

projectId (optional)

`storageCountExperimentsResponse` - Up

count (optional)

String format: int64

`storageCountModelVersionsRequest` - Up

registeredModelId (optional)

`storageCountModelVersionsResponse` - Up

count (optional)

Integer format: int32

`storageCountRegisteredModelRequest` - Up

projectId (optional)

`storageCountRegisteredModelResponse` - Up

registeredModelCount (optional)

Integer format: int32

`storageCreateArtifactRequest` - Up

artifact (optional)

`storageCreateArtifactResponse` - Up

artifact (optional)

`storageCreateDatasetRequest` - Up

dataset (optional)

projectId (optional)

responseMetadata (optional)

`storageCreateDatasetResponse` - Up

dataset (optional)

`storageCreateDeploymentEnvironmentRequest` - Up

deploymentEnvironment (optional)

`storageCreateDeploymentEnvironmentResponse` - Up

deploymentEnvironment (optional)

`storageCreateExperimentCommentRequest` - Up

experimentId (optional)

String ID of the Experiment to comment on.

commentMessage (optional)

String The actual text of the comment.

`storageCreateExperimentCommentResponse` - Up

comment (optional)

storageComment

`storageCreateExperimentMetricRequest` - Up

experimentId (optional)

String ID of the Experiment to add metric to.

datasetId (optional)

String ID of the Dataset the metric relates to.

key (optional)

String Key of the metric unique for the <experiment,dataset> pair.

value (optional)

storageValue

`storageCreateExperimentMetricResponse` - Up

metric (optional)

storageMetric

`storageCreateExperimentRequest` - Up

experiment (optional)

projectId (optional)

responseMetadata (optional)

disableModelRegistration (optional)

Boolean By default, this will be false, and creating an experiment will also trigger model registration. If a client wants to disable model registration, this flag must be set to true.

`storageCreateExperimentResponse` - Up

experiment (optional)

storageRegisteredModelVersion

`storageCreateModelVersionRequest` - Up

registeredModelVersion (optional)

registeredModelDisplayName (optional)

String Needed when model_id is not provided.

projectId (optional)

String Needed when model_id is not provided.

`storageCreateModelVersionResponse` - Up

registeredModelVersion (optional)

storageRegisteredModelVersion

`storageCreateProjectRequest` - Up

project (optional)

`storageCreateProjectResponse` - Up

project (optional)

`storageCreateRegisteredModelRequest` - Up

registeredModel (optional)

`storageCreateRegisteredModelResponse` - Up

registeredModel (optional)

`storageCreateTagRequest` - Up

tag (optional)

projectId (optional)

`storageCreateTagResponse` - Up

tag (optional)

`storageDataType` - Up

`storageDataValue` - Up

boolValue (optional)

int32Value (optional)

Integer format: int32

int64Value (optional)

String format: int64

float32Value (optional)

Float format: float

float64Value (optional)

Double format: double

datetime64Value (optional)

String Time value as unix timestamp in millis. format: int64

strValue (optional)

`storageDataset` - Up

Represents a single dataset, i.e., tabular data to run training, fitting, validation, and testing on.

id (optional)

String An ID of the entity, unique across all entity types.

displayName (optional)

String User readable name of the entity to be displayed in the UIs.

ownerId (optional)

String ID of the user owning this entity.

createdTime (optional)

Date Creation timestamp. format: date-time

lastModifiedTime (optional)

Date Timestamp of last modification. format: date-time

metadata (optional)

rowCount (optional)

String Number of rows in the dataset. format: int64

colCount (optional)

String Number of columns in the dataset. format: int64

size (optional)

String The actual uncompressed dataset file size in bytes. format: int64

statistics (optional)

storageDatasetStats

tag (optional)

array[storageTag] Tags this dataset is annotated with.

`storageDatasetStats` - Up

column (optional)

array[storageColumn]

`storageDeleteArtifactRequest` - Up

id (optional)

`storageDeleteDatasetRequest` - Up

id (optional)

projectId (optional)

`storageDeleteDatasetResponse` - Up

id (optional)

status (optional)

message (optional)

projectId (optional)

`storageDeleteDeploymentEnvironmentRequest` - Up

deploymentEnvironmentId (optional)

`storageDeleteExperimentRequest` - Up

id (optional)

projectId (optional)

`storageDeleteExperimentResponse` - Up

id (optional)

status (optional)

message (optional)

projectId (optional)

`storageDeleteModelVersionRequest` - Up

id (optional)

`storageDeleteModelVersionResponse` - Up

id (optional)

status (optional)

message (optional)

`storageDeleteProjectRequest` - Up

projectId (optional)

`storageDeleteProjectResponse` - Up

projectId (optional)

String ID of the project.

status (optional)

Boolean Status of the delete operation - true or false.

message (optional)

String Error message if the status is false.

`storageDeleteRegisteredModelRequest` - Up

modelId (optional)

`storageDeleteRegisteredModelResponse` - Up

modelId (optional)

String ID of the model that was sent for deletion.

status (optional)

Boolean Status of the delete operation.

message (optional)

String In case the operation failed, this field would have the message.

`storageDeleteTagRequest` - Up

id (optional)

projectId (optional)

`storageDeployRequest` - Up

experimentId (optional)

deploymentEnvironmentId (optional)

array[storageSecondaryScorer]

type (optional)

storageDeploymentType

secondaryScorer (optional)

metadata (optional)

responseMetadata (optional)

deploymentId (optional)

deployerDataVersion (optional)

String These fields are dedicated for the Deployer service to persist arbitrary data. Read and write of these fields is bound to extra permissions.

deployerData (optional)

byte[] format: byte

displayName (optional)

`storageDeployResponse` - Up

deployment (optional)

array[storageSecondaryScorer]

`storageDeployment` - Up

Represents a deployment of a model scorer.

id (optional)

String An ID of the entity, unique across all entity types.

createdTime (optional)

Date Creation timestamp. format: date-time

lastModifiedTime (optional)

Date Last modification timestamp. format: date-time

projectId (optional)

String ID of the project the deployment is part of.

deploymentEnvironmentId (optional)

String ID of the deployment environment, i.e., a template for the actual deployment.

experimentId (optional)

String ID of the experiment whose model is the primary model of this deployment. There can be additional models handling a fraction or a copy of the traffic, e.g., as a part of A/B testing or Champion/Challenger scenario. See field secondary_scorer for these.

type (optional)

storageDeploymentType

secondaryScorer (optional)

Additional scorers defined for the deployment. Lists experiment whose scorer should also receive and/or handle a fraction or copy of the traffic.

Note that all the experiments have to belong to the same project identified by project_id.

metadata (optional)

deployerDataVersion (optional)

String These fields are dedicated for the Deployer service to persist arbitrary data. Read and write of these fields is bound to extra permissions.

deployerData (optional)

byte[] format: byte

ownerId (optional)

String Owner of the deployment.

displayName (optional)

`storageDeploymentEnvironment` - Up

Represents a deployment environment, i.e., a template for deploying a trained model scorer.

id (optional)

String An ID of the entity, unique across all entity types.

displayName (optional)

String User readable name of the entity to be displayed in the UIs.

createdTime (optional)

Date Creation timestamp. format: date-time

lastModifiedTime (optional)

Date Last modification timestamp. format: date-time

projectId (optional)

String ID of the project the deployment environment is part of.

deploymentTargetName (optional)

Unique name of the target environment as defined in Deployer's configuration file.

In the simple cases, it is likely that the Deployer would only have a single K8s environment configured and this would always specify that particular K8s environment. In general, however, we want to support multiple target environments, e.g., to support running development scorers in AWS and production scorer's in on-premise K8s.

`storageDeploymentType` - Up

SINGLE_MODEL: Only the primary model of the deployment is handling the traffic, no secondary scorers are allowed.
SHADOW_TRAFFIC: The primary model of the deployment is handling all the traffic, but secondary scorers are specified. The additional scorers are deployed and receive a copy of the traffic, but their predictions are not used.

This deployment type is useful to implement the Champion/Challenger scenario. The primary model scorer, the champion, is used to handle all the requests. However, the new candidate scorer(s), the challenger, is still supplied all the requests and its predictions are being monitored for later evaluation, e.g., as data to sign off replacing the champion, but not actually used in the scoring response.

SPLIT_TRAFFIC: Traffic is split between multiple models with the primary model receiving the remaining fraction of the traffic after all the secondary scorers are processed. For example, if there are two secondary scorers with fractions 0.1 (10%) and 0.2 (20%), the primary model is handling 0.7 (70%) of the traffic.

This routing is useful to implement A/B testing, where additional models being tested handle a gradually increasing fraction of the traffic. In contrast with SHADOW_TRAFFIC, the predictions of secondary models are actually used in the scoring responses in the corresponding fractions.

`storageEntityType` - Up

`storageExperiment` - Up

Represents a single ML experiment, its metadata, and any artifacts.

id (optional)

String An ID of the entity, unique across all entity types.

displayName (optional)

String User readable name of the entity to be displayed in the UIs.

ownerId (optional)

String ID of the user owning this entity.

createdTime (optional)

Date Creation timestamp. format: date-time

lastModifiedTime (optional)

Date Timestamp of last modification. format: date-time

metadata (optional)

storageExperimentParameters

parameters (optional)

status (optional)

ExperimentExperimentStatus

statistics (optional)

storageExperimentStatistics

tag (optional)

array[storageTag] Tags this experiment is annotated with.

description (optional)

String Description of the experiment.

`storageExperimentInfo` - Up

experiment (optional)

modelInfo (optional)

storageModelInfo

`storageExperimentParameters` - Up

Original configuration parameters of the experiment.

trainingDatasetId (optional)

String If set, contains ID of the dataset used for training.

validationDatasetId (optional)

String If set, contains ID of the dataset used for validation.

testDatasetId (optional)

String If set, contains ID of the dataset used for testing.

targetColumn (optional)

String The column to predict.

weightColumn (optional)

String Column that indicates the observation weight (a.k.a. sample or row weight), if applicable.

foldColumn (optional)

String Column to use to create stratification folds during (cross-)validation, if applicable.

`storageExperimentStatistics` - Up

Statistics about the experiment progress & result.

trainingDuration (optional)

String Duration of model training.

`storageFilterRequest` - Up

Generic filter request specifying criteria on the request results.

query (optional)

storageQuery

`storageGetArtifactRequest` - Up

id (optional)

`storageGetArtifactResponse` - Up

artifact (optional)

`storageGetCommentRequest` - Up

id (optional)

`storageGetCommentResponse` - Up

comment (optional)

storageComment

`storageGetDatasetRequest` - Up

id (optional)

responseMetadata (optional)

`storageGetDatasetResponse` - Up

dataset (optional)

`storageGetDeploymentEnvironmentRequest` - Up

deploymentEnvironmentId (optional)

`storageGetDeploymentEnvironmentResponse` - Up

deploymentEnvironment (optional)

`storageGetDeploymentRequest` - Up

deploymentId (optional)

responseMetadata (optional)

`storageGetDeploymentResponse` - Up

deployment (optional)

`storageGetExperimentRequest` - Up

id (optional)

responseMetadata (optional)

`storageGetExperimentResponse` - Up

experiment (optional)

`storageGetModelVersionForExperimentRequest` - Up

experimentId (optional)

filter (optional)

array[storageRegisteredModelVersion]

`storageGetModelVersionForExperimentResponse` - Up

registeredModelVersion (optional)

`storageGetModelVersionRequest` - Up

modelVersionId (optional)

storageRegisteredModelVersion

`storageGetModelVersionResponse` - Up

registeredModelVersion (optional)

`storageGetProjectRequest` - Up

projectId (optional)

`storageGetProjectResponse` - Up

project (optional)

`storageGetProjectWithAggregatedInfoRequest` - Up

projectId (optional)

`storageGetProjectWithAggregatedInfoResponse` - Up

projectInfo (optional)

storageProjectInfo

`storageGetRegisteredModelByNameRequest` - Up

modelName (optional)

projectId (optional)

`storageGetRegisteredModelByNameResponse` - Up

registeredModel (optional)

`storageGetRegisteredModelRequest` - Up

modelId (optional)

`storageGetRegisteredModelResponse` - Up

registeredModel (optional)

`storageGetTagRequest` - Up

id (optional)

projectId (optional)

`storageGetTagResponse` - Up

tag (optional)

`storageGetUserRequest` - Up

id (optional)

`storageGetUserResponse` - Up

user (optional)

storageUser

`storageHistogramEntry` - Up

tick (optional)

String TODO(osery): Should these be of the correct column type? Maybe use the DataValue type here.

count (optional)

String format: int64

`storageInclusion` - Up

Types of inclusion constraints.

INCLUDE: Must include constraint (default value if unspecified).
EXCLUDE: Must not include constraint.

`storageIsExperimentRegisteredRequest` - Up

experimentId (optional)

projectId (optional)

`storageIsExperimentRegisteredResponse` - Up

isRegistered (optional)

`storageKeySelection` - Up

Identifies the a set of metadata or metric keys (or key patterns) to include in a response. There is an implicit OR between the patterns and no patterns means don't include anything (default).

pattern (optional)

array[String] Key name or key pattern (if including %) to include. The % has the standard SQL LIKE meaning and escaping (i.e., % means any possibly empty substring and %% means just the character %).

`storageLinkDatasetIntoProjectRequest` - Up

projectId (optional)

datasetId (optional)

`storageLinkExperimentIntoProjectRequest` - Up

projectId (optional)

experimentId (optional)

disableModelRegistration (optional)

Boolean By default, this will be false, and linking an experiment will also trigger model registration. If a client wants to disable model registration, this flag must be set to true.

`storageListAllDeploymentsRequest` - Up

paging (optional)

filter (optional)

sorting (optional)

responseMetadata (optional)

`storageListAllDeploymentsResponse` - Up

deployment (optional)

array[storageDeployment]

paging (optional)

`storageListAllProjectsRequest` - Up

paging (optional)

filter (optional)

sorting (optional)

`storageListAllProjectsResponse` - Up

project (optional)

array[storageProject]

paging (optional)

totalProjectCount (optional)

String format: int64

`storageListDatasetsRequest` - Up

paging (optional)

filter (optional)

sorting (optional)

projectId (optional)

responseMetadata (optional)

`storageListDatasetsResponse` - Up

paging (optional)

dataset (optional)

array[storageDataset]

`storageListDeploymentEnvironmentsRequest` - Up

projectId (optional)

paging (optional)

filter (optional)

sorting (optional)

`storageListDeploymentEnvironmentsResponse` - Up

paging (optional)

array[storageDeploymentEnvironment]

deploymentEnvironment (optional)

`storageListDeploymentsRequest` - Up

projectId (optional)

paging (optional)

filter (optional)

sorting (optional)

responseMetadata (optional)

`storageListDeploymentsResponse` - Up

paging (optional)

deployment (optional)

array[storageDeployment]

`storageListEntityArtifactsRequest` - Up

paging (optional)

filter (optional)

sorting (optional)

entityId (optional)

String ID of the parent entity (required).

entityType (optional)

storageEntityType

`storageListEntityArtifactsResponse` - Up

paging (optional)

artifact (optional)

array[storageArtifact]

`storageListExperimentCommentsRequest` - Up

experimentId (optional)

String ID of the experiment whose comments should be listed.

paging (optional)

filter (optional)

sorting (optional)

`storageListExperimentCommentsResponse` - Up

paging (optional)

comment (optional)

array[storageComment]

`storageListExperimentMetricsRequest` - Up

experimentId (optional)

paging (optional)

filter (optional)

sorting (optional)

`storageListExperimentMetricsResponse` - Up

paging (optional)

metric (optional)

array[storageMetric]

`storageListExperimentsRequest` - Up

paging (optional)

filter (optional)

sorting (optional)

projectId (optional)

responseMetadata (optional)

`storageListExperimentsResponse` - Up

paging (optional)

experiment (optional)

array[storageExperiment]

`storageListExperimentsWithModelInfoRequest` - Up

projectId (optional)

paging (optional)

filter (optional)

sorting (optional)

responseMetadata (optional)

array[storageExperimentInfo]

`storageListExperimentsWithModelInfoResponse` - Up

experimentInfo (optional)

paging (optional)

experimentCount (optional)

String format: int64

`storageListModelVersionsForModelRequest` - Up

registeredModelId (optional)

paging (optional)

filter (optional)

sorting (optional)

array[storageRegisteredModelVersion]

`storageListModelVersionsForModelResponse` - Up

modelVersions (optional)

paging (optional)

`storageListProjectEventsRequest` - Up

projectId (optional)

paging (optional)

filter (optional)

sorting (optional)

`storageListProjectEventsResponse` - Up

event (optional)

array[storageAuditEvent]

paging (optional)

eventCount (optional)

String format: int64

`storageListProjectSharingsRequest` - Up

projectId (optional)

`storageListProjectSharingsResponse` - Up

sharing (optional)

array[storageSharing]

`storageListProjectsContainingExperimentRequest` - Up

experimentId (optional)

paging (optional)

filter (optional)

sorting (optional)

`storageListProjectsContainingExperimentResponse` - Up

project (optional)

array[storageProject]

paging (optional)

`storageListProjectsRequest` - Up

paging (optional)

filter (optional)

sorting (optional)

`storageListProjectsResponse` - Up

project (optional)

array[storageProject]

paging (optional)

userProjectCount (optional)

String format: int64

`storageListProjectsWithAggregatedInfoRequest` - Up

paging (optional)

filter (optional)

sorting (optional)

array[storageProjectInfo]

`storageListProjectsWithAggregatedInfoResponse` - Up

projectInfo (optional)

paging (optional)

userProjectCount (optional)

String format: int64

`storageListRegisteredModelsRequest` - Up

projectId (optional)

String ID of the project from which the listing has to be happen. Users with permission LIST_ALL_REGISTERED_MODELS can leave this empty to list models across all projects.

paging (optional)

filter (optional)

sorting (optional)

array[storageRegisteredModel]

`storageListRegisteredModelsResponse` - Up

registeredModels (optional)

paging (optional)

`storageListRolesResponse` - Up

role (optional)

array[storageRole]

`storageListTagsRequest` - Up

paging (optional)

filter (optional)

sorting (optional)

projectId (optional)

`storageListTagsResponse` - Up

paging (optional)

tag (optional)

array[storageTag]

`storageListUsersRequest` - Up

paging (optional)

filter (optional)

sorting (optional)

`storageListUsersResponse` - Up

paging (optional)

map[String, storageValue]

user (optional)

array[storageUser]

`storageMetadata` - Up

Represents a generic holder for different ML backend specific key value pairs. The keys should identify the ML backend (and where relevant for backward/forward compatibility also a version) such as, e.g, "dai/resumed_experiment" or "dai:1.5.1/model_parameters".

values (optional)

`storageMetadataReference` - Up

key (optional)

String Key of the metadata value.

`storageMetric` - Up

Represents a single metric or possibly other metadata about experiment/model bound to an input dataset, e.g., the training or validation dataset.

Specific ML backends can also store proprietary scoring information using the json_value or proto_value for the given <experiment,dataset> pair. Information specific to just an experiment is stored in experiment metadata instead.

experimentId (optional)

String ID of the experiment this metric belongs to.

datasetId (optional)

String ID of the Dataset this metric was based on.

value (optional)

storageValue

key (optional)

String Uniquely identifies metric for the <experiment,dataset> combination.

`storageMetricReference` - Up

datasetReference (optional)

MetricReferenceDatasetReference

datasetId (optional)

String Id of a custom dataset. Only used when dataset_reference == CUSTOM_DATASET.

key (optional)

String Key of the metric value.

`storageModelInfo` - Up

modelId (optional)

modelName (optional)

modelDescription (optional)

projectId (optional)

modelOwner (optional)

modelVersionId (optional)

modelVersionNumber (optional)

modelVersionOwner (optional)

`storageNonNumericColumnStats` - Up

top (optional)

String TODO(osery): Should this be of the correct column type? Maybe use the DataValue type here.

`storageNumericColumnStats` - Up

mean (optional)

Double format: double

std (optional)

Double format: double

min (optional)

Double format: double

max (optional)

Double format: double

`storageOperation` - Up

DOWNLOAD_ARTIFACT: --- Artifact operations ---
GET_PROJECT: Project operations.
GET_ARTIFACT: Artifact operations.
GET_COMMENT: Comment operations.
GET_DATASET: Dataset operations.
GET_DEPLOYMENT_ENVIRONMENT: Deployment Environment operations.
GET_DEPLOYMENT: Deployment operations.
GET_EXPERIMENT: Experiment operations.
CREATE_TAG: Tag operations.
CREATE_REGISTERED_MODEL: Model operations.
GET_REGISTERED_MODEL_VERSION: ModelVersion operation.
GET_ENDPOINT: Endpoint operations.
GET_BATCH_SCORING_JOB: Batch scoring operations.

`storageOperator` - Up

Supported types of comparison operators.

EQUAL_TO: Operands are equal, corresponding to textual '=' or '=='.
NOT_EQUAL_TO: Operands are not equal, corresponding to textual '!=' or '<>'.
LESS_THAN: First operand is strictly lesser than the second, corresponding to textual '<'.
LESS_OR_EQUAL_TO: First operand is lesser than or equal to the second, corresponding to textual '<='.
GREATER_THAN: First operand is strictly greater than the second, corresponding to textual '>'.
GREATER_OR_EQUAL_TO: First operand is greater than or equal to the second, corresponding to textual '>='.
CONTAINS: First operand contains the second (e.g., a substring).
ICONTAINS: First operand contains the second (e.g., a substring) ignoring case.

`storageOrder` - Up

Types of field ordering for listing requests.

ASCENDING: Sort records by the given field in an ascending order natural for its type (default value).
DESCENDING: Sort records by the given field in a descending order natural for its type.

`storagePagingRequest` - Up

Generic paging request portion needed for listing in presence of concurrent modifications.

pageSize (optional)

Integer Requested number of records to be returned. Note that fewer records may be returned due to server-side restrictions or simply if there are no more records in on the server. format: int32

pageToken (optional)

byte[] Use PagingResponse.next_page_token here to continue with the next record under any given filter and sorting. format: byte

`storagePagingResponse` - Up

Generic paging response portion needed for listing in presence of concurrent modifications.

nextPageToken (optional)

byte[] An opaque token to be sent together with the next listing request to continue with the next record under any given filter and sorting. This token should not be interpreted by the client and its structure may change without any prior notice. format: byte

`storageProject` - Up

id (optional)

String An ID of the entity, unique across all entity types.

displayName (optional)

String User readable name of the entity to be displayed in the UIs.

ownerId (optional)

String ID of the user owning this entity.

createdTime (optional)

Date Creation timestamp. format: date-time

lastModifiedTime (optional)

Date Timestamp of last modification. format: date-time

description (optional)

String User provided textual description of the project.

`storageProjectInfo` - Up

Represents the aggregated information about a project.

project (optional)

registeredModelCount (optional)

Integer format: int32

shareCount (optional)

Integer Count of the number of users the project is shared with as provided by the SharingService. format: int32

experimentCount (optional)

String Count of the number of experiments within the project as provided by the ExperimentService. format: int64

sharedUser (optional)

array[storageUser] The information with regards to the users with whom the project has been shared with, A maximum of 10 user details are provided as part of the response.

registeredModelVersionCount (optional)

Integer Count of the number of model versions across all registered models in the project. format: int32

`storageProperty` - Up

Specifies a property of an entity, i.e., direct field, metadata, or a metric value.

field (optional)

String Name of a direct field of the entity, e.g., 'display_name'.

metadata (optional)

storageMetadataReference

metric (optional)

storageMetricReference

`storagePropertyConstraint` - Up

property (optional)

storageProperty

operator (optional)

storageOperator

value (optional)

storageValue

`storageQuery` - Up

Disjunction of multiple constraints.

clause (optional)

array[storageClause]

There is an implicit disjunction, OR, between the query clauses.

WARNING: Note that we currently don't support disjunction of clauses and an error will be returned for requests containing multiple clauses.

`storageRecoverDeletedExperimentRequest` - Up

experimentId (optional)

projectId (optional)

`storageRecoverDeletedExperimentResponse` - Up

experimentId (optional)

projectId (optional)

status (optional)

message (optional)

`storageRegisteredModel` - Up

id (optional)

String An ID of the registered model.

displayName (optional)

description (optional)

projectId (optional)

createdTime (optional)

Date format: date-time

createdBy (optional)

String ID of the user owning this model.

updatedTime (optional)

Date format: date-time

updatedBy (optional)

String ID of the user last update this model.

status (optional)

`storageRegisteredModelVersion` - Up

id (optional)

registeredModelId (optional)

experimentId (optional)

storageRegisteredModelVersionStatus

version (optional)

Integer format: int32

status (optional)

createdTime (optional)

Date format: date-time

createdBy (optional)

updatedTime (optional)

Date format: date-time

updatedBy (optional)

`storageRegisteredModelVersionStatus` - Up

`storageRole` - Up

Represents a single sharing record for an entity.

id (optional)

String Unique ID of the role record.

displayName (optional)

String Display name of the role.

createdTime (optional)

Date Creation timestamp. format: date-time

`storageSecondaryScorer` - Up

Defines a secondary model that handles a fraction or a copy of the traffic.

experimentId (optional)

array[storageSortProperty]

fraction (optional)

Double format: double

`storageShareProjectRequest` - Up

projectId (optional)

String ID of the project being shared.

userId (optional)

String If non-empty, an ID of a user to which the access is granted. Cannot be set at the same time as group.

restrictionRoleId (optional)

String If non-empty, restricts the allowed actions resulting from this sharing record to at most actions allowed by the specified role.

`storageShareProjectResponse` - Up

sharing (optional)

storageSharing

`storageSharing` - Up

Represents a single sharing record for an entity.

id (optional)

String Unique ID of the sharing record.

entityId (optional)

String ID of an entity that is being shared.

rootSharingId (optional)

String For project entities, this field is always empty. For other entities, this field contains the ID of the project sharing record, where the sharing of the entity originates.

userId (optional)

String If non-empty, an ID of a user to which the access is granted. Cannot be set at the same time as group.

restrictionRoleId (optional)

String If non-empty, restricts the allowed actions resulting from this sharing record to at most actions allowed by the specified role.

type (optional)

storageSharingType

createdTime (optional)

Date Creation timestamp. format: date-time

`storageSharingType` - Up

USER_CREATED: Sharing directly created by a user. A user created sharing must have an empty root_sharing_id and can be manipulated via the API.
OWNER: Sharing representing the owner's implicit sharing. An owner sharing must have an empty root_sharing_id but cannot be manipulated via the API.
IMPLIED: Sharing implied by another sharing. An implied sharing must have its root_sharing_id set to the root sharing, either USER_CREATED or OWNER.

`storageSortProperty` - Up

Defines a single property to sort requests results on together with the desired ordering direction.

property (optional)

storageProperty

order (optional)

storageOrder

`storageSortingRequest` - Up

Generic sorting request specifying the ordering of a request results from the most to the least important properties.

property (optional)

List of sorting properties in the decreasing order of importance.

WARNING: Note that we currently don't support multiple sort properties.

`storageTag` - Up

Tag is a way to mark entities. Tags can be either unique or multi tags. Unique tags can only mark a single entity. Multi tags can mark multiple entities at once.

id (optional)

String An ID of the entity, unique across all entity types.

displayName (optional)

String User readable name of the entity to be displayed in the UIs.

createdTime (optional)

Date Creation timestamp. format: date-time

projectId (optional)

String ID of the parent project entity. Currently, only project-scoped tags are supported.

`storageTagConstraint` - Specifies a constraint for a single tag Up

tagId (optional)

String Id of an entity tag.

inclusion (optional)

storageInclusion

`storageTagDatasetRequest` - Up

datasetId (optional)

String ID of the Dataset to tag.

tagId (optional)

String ID of the Tag to assign to the Dataset. Note that if this is a unique tag, the tag would be atomically removed from the previous entity, if already used.

`storageTagExperimentRequest` - Up

experimentId (optional)

String ID of the Experiment to tag.

tagId (optional)

String ID of the Tag to assign to the Experiment. Note that if this is a unique tag, the tag would be atomically removed from the previous entity, if already used.

`storageTimeColumnStats` - Up

format (optional)

min (optional)

String Minimal time as unix timestamp in millis. format: int64

max (optional)

String Maximal time as unix timestamp in millis. format: int64

`storageTransferProjectOwnershipRequest` - Up

projectId (optional)

newOwnerId (optional)

`storageTransferProjectOwnershipResponse` - Up

projectId (optional)

ownerId (optional)

status (optional)

`storageUndeployRequest` - Up

deploymentId (optional)

`storageUnlinkDatasetFromProjectRequest` - Up

projectId (optional)

datasetId (optional)

`storageUnlinkExperimentFromProjectRequest` - Up

projectId (optional)

experimentId (optional)

`storageUnshareProjectRequest` - Up

sharingId (optional)

`storageUntagDatasetRequest` - Up

datasetId (optional)

String ID of the Dataset to untag.

tagId (optional)

String ID of the Tag to remove from the Dataset.

`storageUntagExperimentRequest` - Up

experimentId (optional)

String ID of the Experiment to untag.

tagId (optional)

String ID of the Tag to remove from the Experiment.

`storageUpdateArtifactRequest` - Up

artifact (optional)

updateMask (optional)

`storageUpdateArtifactResponse` - Up

artifact (optional)

`storageUpdateDatasetRequest` - Up

dataset (optional)

updateMask (optional)

responseMetadata (optional)

`storageUpdateDatasetResponse` - Up

dataset (optional)

`storageUpdateDeploymentEnvironmentRequest` - Up

deploymentEnvironment (optional)

updateMask (optional)

`storageUpdateDeploymentEnvironmentResponse` - Up

deploymentEnvironment (optional)

`storageUpdateDeploymentRequest` - Up

deployment (optional)

updateMask (optional)

responseMetadata (optional)

`storageUpdateDeploymentResponse` - Up

deployment (optional)

`storageUpdateExperimentRequest` - Up

experiment (optional)

updateMask (optional)

responseMetadata (optional)

`storageUpdateExperimentResponse` - Up

experiment (optional)

`storageUpdateProjectRequest` - Up

project (optional)

updateMask (optional)

`storageUpdateProjectResponse` - Up

project (optional)

`storageUpdateRegisteredModelRequest` - Up

registeredModel (optional)

updateMask (optional)

`storageUpdateRegisteredModelResponse` - Up

registeredModel (optional)

`storageUpdateTagRequest` - Up

tag (optional)

updateMask (optional)

`storageUpdateTagResponse` - Up

tag (optional)

`storageUser` - Up

id (optional)

username (optional)

primaryEmail (optional)

createdTime (optional)

Date Creation timestamp. format: date-time

idpId (optional)

String Deprecated: Migrate to id.

`storageValue` - Up

Represents a generic holder for different value types for metrics and parameters.

doubleValue (optional)

Double format: double

stringValue (optional)

int64Value (optional)

String format: int64

boolValue (optional)

jsonValue (optional)

timestampValue (optional)

Date format: date-time

durationValue (optional)

protoValue (optional)

protobufAny

idValue (optional)

`v1BatchParameters` - Up

Configuration parameters for controlling batch processing behavior during model inference, including batch size optimization settings.

miniBatchSize (optional)

Integer Optional. The number of records to send in one request to the scorer. If not provided, 64 will be used as default. This value cannot be modified after job creation due to the IMMUTABLE constraint. Larger batch sizes can improve throughput but may increase memory usage. format: int32

`v1CancelJobResponse` - Response message for CancelJob Up

job (optional)

v1Job

`v1CompletionStats` - Up

CompletionStats describe batch-scoring job completion statistics after it has been completed or terminated.

processedRows (optional)

String Required. Number of processed input input when batch soring job is completed. format: int64

finishedPipelineCount (optional)

Integer Required. Number of finished pipeline count. format: int32

failedPipelines (optional)

v1KubernetesConfigShortcut

errorRowsCount (optional)

String Required. Number of rows that failed scoring. format: int64

`v1CreateJobResponse` - Response message for CreateJob Up

job (optional)

v1Job

`v1DeploymentComposition` - Up

DeploymentComposition represents a deployment composition of a model.

experimentId

String Required. ID of the experiment hosting the model to be deployed.

artifactType

String Required. Identifying name of the deployable artifact type.

artifactProcessor

String Required. Identifying name of the artifact processor.

runtime

String Required. Identifying name of the runtime.

runtimeImage (optional)

String Optional. If specified, runtime image will be overwritten with this value if allowed.

`v1GetJobResponse` - Response message for GetJob Up

job (optional)

v1Job

`v1GetSinkSpecResponse` - Response message for GetSinkSpec Up

sinkSpec (optional)

v1SinkSpec

`v1GetSourceSpecResponse` - Response message for GetSourceSpec Up

sourceSpec (optional)

v1SourceSpec

`v1InstanceSpec` - Up

Specification for configuring a deployed instance, including its composition, resource allocation, environment, and Kubernetes settings.

deploymentComposition

v1DeploymentComposition

resourceSpec (optional)

v1ResourceSpec

environmentVariables (optional)

map[String, String] Optional. The environment variables to set in the scorer runtime.

kubernetesConfigShortcut (optional)

`v1Job` - Job is entity representing Batch Scoring Job Up

name (optional)

String Resource name.

displayName (optional)

String Optional. Human-readable name.

source

v1Source

sink

v1Sink

instanceSpec

v1InstanceSpec

batchParameters (optional)

v1BatchParameters

modelRequestParameters (optional)

v1ModelRequestParameters

state (optional)

JobState

creator (optional)

String Output only. Name of an entity that created the Job.

createTime (optional)

Date Output only. Time when the job was created. format: date-time

startTime (optional)

Date Time when the Job started executing. format: date-time

endTime (optional)

Date Time when the Job completed. format: date-time

completionStats (optional)

v1CompletionStats

jobTimeout (optional)

String Optional. Maximum allowed time for the job to run before being automatically timed out. Example values in JSON: "15s", "10m", "1h".

`v1JobOutputResponse` - Up

Response message for GetJobOutput.

appType (optional)

ModelRequestParametersShapleyType

pod (optional)

String The name of the Kubernetes pod that produced this log entry.

container (optional)

String The name of the container within the pod that generated this output.

logTime (optional)

Date The timestamp when this log entry was generated. format: date-time

line (optional)

String The actual log line or output content from the application.

error (optional)

String Error message if there was an issue retrieving or processing the log output. Empty if no error occurred.

`v1KubernetesConfigShortcut` - Up

Shortcut configuration options that translate to expanded Kubernetes resource configuration.

kubernetesAffinityShortcut (optional)

String Optional. The name of the KubernetesAffinityShortcut to apply to the batch-scoring job.

kubernetesTolerationShortcut (optional)

String Optional. The name of the KubernetesTolerationShortcut to apply to the batch-scoring job.

`v1ListJobsResponse` - Response message for ListJobs Up

jobs (optional)

array[v1Job] Collection of jobs.

nextPageToken (optional)

String Used to retrieve the next page of results. When unset, no further items are available (this response was the last page).

`v1ListSinkSpecsResponse` - Response message for ListSinkSpecs (-- api-linter: core::0158::response-next-page-token-field=disabled aip.dev/not-precedent: There will be only couple of specs. --) Up

sinkSpecs (optional)

array[v1SinkSpec] Collection of specs.

`v1ListSourceSpecsResponse` - Response message for ListSourceSpecs (-- api-linter: core::0158::response-next-page-token-field=disabled aip.dev/not-precedent: There will be only couple of specs. --) Up

sourceSpecs (optional)

array[v1SourceSpec] Collection of specs.

`v1ModelRequestParameters` - Up

Parameters for configuring model inference requests, including feature contributions, prediction intervals, and record identification options.

idField (optional)

String Optional. Field name to use as a unique identifier for each prediction record. When specified, this field's value will be included in the response to help correlate predictions with input data.

requestContributions (optional)

requestPredictionIntervals (optional)

Boolean Optional. Whether to include prediction intervals in the model response. When true, the response will contain confidence intervals or uncertainty estimates for the predictions in addition to point predictions.

`v1ResourceRequirement` - Up

ResourceRequirement describes the scorer container resource requirements.

requests (optional)

map[String, String] Optional. Resource requests for the scorer container. If requests is omitted, it defaults to limits if that is explicitly specified, otherwise to an implementation-defined value.

limits (optional)

map[String, String] Optional. Resource limits for the scorer container. If not specified, there are no limits.

`v1ResourceSpec` - Up

ResourceSpec describes the resource specification for the pod.

resourceRequirement (optional)

v1ResourceRequirement

replicas (optional)

Integer Optional. Number of replicas for the scorer container. Defaults to 1. format: int32

minimalAvailableReplicas (optional)

Integer format: int32

`v1Sink` - Up

Configuration for a data sink destination where batch processing results will be written. All fields are immutable after creation.

spec

String Required. Reference to the SinkSpec resource that defines the sink type and capabilities (e.g., file system, database, cloud storage). Cannot be changed after the sink is created.

config

String Required. Configuration parameters specific to the sink type, such as connection strings, authentication details, or format specifications. Cannot be modified after the sink is created.

location

String Required. The target location or path where data will be written, such as a file path, database table name, or cloud storage URI. Cannot be changed after the sink is created.

mimeType (optional)

Specifies the mime type for file-based sinks. This value determines the format used when writing data to the output files. Common mime types include:

text/csv: For CSV (Comma-Separated Values) format, used for structured tabular data.
application/jsonl: For JSON Lines (newline-delimited JSON) format, where each line is a separate JSON object.
application/json: For standard JSON format, used for structured data in a hierarchical or nested structure.

Specifying the correct mime type is essential to ensure data is output in the desired format. If not specified, the system may apply a default format that may not meet expectations.

`v1SinkSpec` - Up

Specification defining the capabilities and schema for a data sink type. This resource describes how to connect to and write data to different types of destinations such as databases, file systems, or cloud storage services.

name (optional)

String Resource name.

displayName (optional)

String Human-readable name.

schema (optional)

String Schema for sink spec configuration. Represented in JSON-schema. It should contains all available parameters for particular connector. To represent secret parameters implementation should return writeOnly annotation for that field.

supportedMimeTypes (optional)

supportedLocationPaths (optional)

`v1Source` - Up

Configuration for a data source from which batch processing input data will be read. All fields are immutable after creation.

spec

String Required. Reference to the SourceSpec resource that defines the source type and capabilities (e.g., file system, database, cloud storage). Cannot be changed after the sink is created.

config

String Required. Configuration parameters specific to the source type, such as connection strings, authentication details, or format specifications. Cannot be modified after the source is created.

location

String Required. The location or path from which data will be read, such as a file path, database table name, or cloud storage URI. Cannot be changed after the source is created.

mimeType (optional)

Specifies the mime type for file-based sources. This value is used to determine how the file data should be interpreted and processed. Common mime types include:

text/csv: For CSV (Comma-Separated Values) files.
application/jsonl: For JSON Lines (newline-delimited JSON) format, typically used for structured data with one JSON object per line.
application/json: For regular JSON files containing structured data.
image/*: For image files of any format (e.g., image/png, image/jpeg).

Providing an appropriate mime type is crucial for the correct handling of file data. If not specified, default behavior may apply.

`v1SourceSpec` - Up

Specification defining the capabilities and schema for a data source type. This resource describes how to connect to and read from different types of data sources such as databases, file systems, or cloud storage services.

name (optional)

String Resource name.

displayName (optional)

String Human-readable name.

schema (optional)

String Schema for source spec configuration. Represented in JSON-schema. It should contains all available parameters for particular connector. To represent secret parameters implementation should return writeOnly annotation for that field.

supportedMimeTypes (optional)

supportedLocationPaths (optional)

AuthenticationPassphraseHashAlgorithm

`v2ArtifactComposition` - Up

ArtifactComposition represents a combination of compatible components required for deployment. It specifies which source artifact types can be processed and deployed to specific runtime environments through defined model types.

deployableArtifactType (optional)

v2DeployableArtifactType

artifactProcessor (optional)

v2ArtifactProcessor

modelType (optional)

v2ModelType

runtime (optional)

v2Runtime

`v2ArtifactProcessor` - Represents a processor that can transform an entity represented by a particular DeployableArtifactType into one representable by a particular ModelType. (-- api-linter: core::0123::resource-annotation=disabled --) Up

name (optional)

String Identifying name for the entity.

displayName (optional)

String Human-readable name of the entity.

deployableArtifactType (optional)

String Name of the DeployableArtifactType this processor is compatible with.

modelType (optional)

String Name of the ModelType this processor is compatible with.

image (optional)

String Full name of the container image for the artifact processor.

`v2AuthenticationPassphrase` - Up

A passphrase is expected to be supplied in an Authorization header in the form Authorization: Bearer <passphrase>.

passphrase

String The plaintext passphrase to authenticate with. When algorithm is not plain text, it won't be returned in response.

hashAlgorithm

`v2AuthorizationAccessToken` - Up

An access token is expected to be supplied in an Authorization header in the form of Authorization: Bearer <token>.

authenticationProtocol

AuthorizationAccessTokenAuthorizationProtocolSecurityType

`v2BaselineAggregation` - BaselineAggregation is the entity which represent base aggregates for feature Up

id (optional)

String Baseline Aggregation ID autogenerated.

column

String Column name of the feature.

logicalType

v2LogicalType

numericalAggregate (optional)

v2NumericalAggregate

categoricalAggregate (optional)

v2CategoricalAggregate

missingValues (optional)

v2MissingValues

deploymentId (optional)

isModelOutput (optional)

`v2BaselineSource` - Up

BaselineSource represent location of baseline data.

path

String Baseline dataset path could be a url to storage.

cred

v2Credentials

`v2BatchGetDeploymentStatusesResponse` - Up

The response message for getting BatchGetDeploymentStatus.

deploymentStatuses (optional)

array[v2DeploymentStatus] The retrieved deployment's status.

errors (optional)

map[String, rpcStatus]

`v2BatchGetDeploymentsMetadataResponse` - Up

The response message for getting BatchGetDeploymentsMetadata.

(-- api-linter: core::0231::response-resource-field=disabled aip.dev/not-precedent: We need to do this because plural form of deploymentMetadata here is deploymentsMetadata. --)

deploymentsMetadata (optional)

array[v2DeploymentMetadata] The retrieved deployments' metadata.

errors (optional)

map[String, rpcStatus]

`v2CategoricalAggregate` - Up

Aggregated metrics specific for columns on LogicalType.LOGICAL_TYPE_CATEGORICAL type. Aggregated metrics are calculated for a specific time window.

valueCounts (optional)

map[String, Integer]

Map of counts of each categorical value fall in to the aggregated window.

Key - Categorical value Value - Count of data points for the categorical value

format: int32

`v2Column` - Up

Column represents a feature column.

id (optional)

String ID of the column. Autogenerated by database.

column

logicalType

v2LogicalType

isModelOutput (optional)

v2KubernetesResourceRequirements

`v2ComputeSuggestedScoringResourceRequirementsResponse` - Up

The response message to represent suggested resource.

kubernetesResourceRequirements (optional)

`v2Cors` - Up

Cors represents the user specified CORS configuration for a deployment.

origins

CredentialsAWSCredentials

`v2CreateDeploymentResponse` - Response message for CreateDeployment Up

deployment (optional)

v2Deployment

`v2CreateEndpointResponse` - Up

The response message for creating an endpoint.

endpoint (optional)

v2Endpoint

`v2Credentials` - Up

Credentials is the entity that represent credentials needed to get access to resources.

aws (optional)

azure (optional)

CredentialsAzureCredentials

`v2DeployableArtifactType` - DeployableArtifactType defines a source artifact entity that can be processed for deployment. It contains metadata about the artifact including its type and identifying information. (-- api-linter: core::0123::resource-annotation=disabled --) Up

name (optional)

String Identifying name for the entity.

displayName (optional)

String Human-readable name of the entity.

artifactType (optional)

String Name of the the target artifact type, as recorded by H2O.ai Storage.

`v2Deployment` - Up

Represents a deployment of a model.

name (optional)

String Resource name.

displayName (optional)

String Optional. Human-readable name.

description (optional)

String Optional. Description of the deployment.

etag (optional)

String The etag of the deployment.

createTime (optional)

Date Creation timestamp. format: date-time

updateTime (optional)

Date Update timestamp. format: date-time

creator (optional)

String Output only. Name of an entity that created the Deployment.

single (optional)

shadow (optional)

split (optional)

security (optional)

cors (optional)

v2Cors

monitoringOptions (optional)

v2MonitoringOptions

`v2DeploymentComposition` - Up

Represents the composition of a deployment, specifying key attributes required for deploying a model.

experimentId

String ID of the experiment hosting the model to be deployed.

deployableArtifactType

String Identifying name of the deployable artifact type.

artifactProcessor

String Identifying name of the artifact processor.

runtime

String Identifying name of the runtime.

runtimeImage (optional)

String Optional. If specified, runtime image will be overwritten with this value if allowed.

`v2DeploymentMetadata` - Up

Represents the metadata of a deployment.

name (optional)

String Resource name.

scoringInterface (optional)

v2ScoringInterface

`v2DeploymentStatus` - Up

Represents the status of a deployment in an environment.

name (optional)

String Resource name.

state (optional)

DeploymentStatusDeploymentState

message (optional)

String Miscellaneous status message.

updateTime (optional)

Date Time of the last status change. format: date-time

`v2DiscoverDefaultRuntimeKubernetesResourceRequirementsResponse` - response for DiscoverDefaultRuntimeKubernetesResourceRequirements Up

defaultRuntimeKubernetesResourceRequirements (optional)

v2KubernetesResourceRequirements

`v2DiscoverKubernetesConfigShortcutsResponse` - Response for DiscoverKubernetesConfigShortcuts Up

kubernetesAffinityShortcuts (optional)

array[v2KubernetesAffinityShortcut] The supported affinity configuration shortcuts.

kubernetesTolerationShortcuts (optional)

array[v2KubernetesTolerationShortcut] The supported toleration configuration shortcuts.

`v2Endpoint` - Up

An endpoint which exposes a target resource.

name (optional)

displayName (optional)

String The human-readable name of this endpoint.

description (optional)

String The description of this endpoint.

path

target (optional)

String If left out or empty, the endpoint is considered detached.

createTime (optional)

Date Output only. The time the endpoint was created. format: date-time

updateTime (optional)

Date Output only. The time the endpoint was last updated. format: date-time

`v2GenerateAggregatorConfigResponse` - Response message for GenerateAggregatorConfig Up

aggregatorConfig (optional)

v2KubernetesResourceRequirements

`v2GenerateLogsResponse` - Up

Response message containing access information for the generated log archive.

artifactId (optional)

String The artifact id to the generated zip file containing logs.

cachedUntilTime (optional)

Date Timestamp until which identical requests will return this same file. Repeated requests before this time will not trigger new log generation. format: date-time

validUntilTime (optional)

Date Timestamp indicating when the log archive will expire. The zip file will be automatically deleted after this time. format: date-time

`v2GetDefaultSecurityOptionTypeResponse` - Response method for GetDefaultSecurityOptionType Up

defaultSecurityOptionType (optional)

v2SecurityOptionType

`v2GetDeploymentMetadataResponse` - Up

The response message for getting an deployment's metadata.

deploymentMetadata (optional)

v2DeploymentMetadata

`v2GetDeploymentResponse` - Up

The response message for getting an deployment.

deployment (optional)

v2Deployment

`v2GetDeploymentStatusResponse` - Up

The response message for getting an deployment's status.

deploymentStatus (optional)

v2DeploymentStatus

`v2GetEndpointResponse` - Up

The response message for getting an endpoint.

endpoint (optional)

v2Endpoint

`v2GetMonitoringConfigResponse` - Response for GetMonitoringConfig Up

monitoringConfig (optional)

v2MonitoringConfig

`v2HttpMethod` - Up

Enumerates different HTTP methods.

HTTP_METHOD_UNSPECIFIED: Unknown method. It should not be used
GET: GET method
POST: POST method

`v2KubernetesAffinityShortcut` - Up

KubernetesAffinityShortcut defines a predefined affinity configuration that can be selected by users during deployment. Each shortcut represents a common affinity pattern that can be applied to Kubernetes workloads without requiring users to understand the full complexity of Kubernetes affinity rules.

This message serves as a descriptor for available affinity shortcuts in the system. When a user selects a shortcut by its name, the Deployer service will apply the corresponding pre-configured affinity rules to the deployment.

Note: The actual affinity configurations are defined server-side and are not part of this message definition, which only serves as a reference for selection. (-- api-linter: core::0123::resource-annotation=disabled --)

name (optional)

String The unique identifying name of the shortcut. This is the value to pass back to the Deployer as part of a request, if this shortcut is to be used.

displayName (optional)

String The human-readable name of this affinity shortcut.

description (optional)

String A description of what the corresponding affinity configuration does.

`v2KubernetesConfigShortcut` - Up

Shortcut configuration options that translate to expanded Kubernetes resource configuration.

kubernetesAffinityShortcut (optional)

String The name of the KubernetesAffinityShortcut to apply to a Kubernetes-bound application.

kubernetesTolerationShortcut (optional)

String The name of the KubernetesTolerationShortcut to apply to a Kubernetes-bound application.

`v2KubernetesResourceRequirements` - Up

Represents compute resource requirements for Kubernetes. Reflects Kubernetes' ResourceRequirements v1 API.

More info: https://kubernetes.io/docs/concepts/configuration/manage-resources-containers/

limits (optional)

map[String, String] Limits describes the maximum amount of compute resources allowed.

requests (optional)

map[String, String] Requests describes the minimum amount of compute resources required. If requests is omitted, it defaults to limits if that is explicitly specified, otherwise to an implementation-defined value.

`v2KubernetesResourceSpec` - Up

Represents Kubernetes resource specifications passed directly to Kubernetes.

kubernetesResourceRequirements (optional)

replicas (optional)

Integer

The number of desired pods for a resulting Kubernetes Deployment.

Reflects the replicas fields of Kubernetes' DeploymentSpec v1 API. no value being set then default will be 1.

format: int32

`v2KubernetesTolerationShortcut` - Up

KubernetesTolerationShortcut defines a predefined toleration configuration that can be selected by users during deployment. Each shortcut represents a common toleration pattern that can be applied to Kubernetes workloads without requiring users to understand the full complexity of Kubernetes toleration rules.

Tolerations allow pods to be scheduled onto nodes with matching taints, which would otherwise repel those pods. This provides a way to control which workloads can run on specific nodes that have been tainted for special purposes (like dedicated hardware, restricted access, or specialized workloads).

This message serves as a descriptor for available toleration shortcuts in the system. When a user selects a shortcut by its name, the Deployer service will apply the corresponding pre-configured toleration rules to the deployment.

Note: The actual toleration configurations are defined server-side and are not part of this message definition, which only serves as a reference for selection. (-- api-linter: core::0123::resource-annotation=disabled --)

name (optional)

String The unique identifying name of the shortcut. This is the value to pass back to the Deployer as part of a request, if this shortcut is to be used.

displayName (optional)

String The human-readable name of this endpoint.

description (optional)

String The description of what the corresponding toleration configuration does.

`v2ListAggregatesResponse` - Response message for ListAggregates Up

aggregates (optional)

array[v2ScoringAggregation]

nextPageToken (optional)

String Used to retrieve the next page of results. When unset, no further items are available (this response was the last page).

`v2ListArtifactCompositionsResponse` - The response message to return list of artifact compositions. (-- api-linter: core::0158::response-next-page-token-field=disabled --) Up

artifactCompositions (optional)

array[v2ArtifactComposition] List of artifact composition entities.

`v2ListBaselineAggregatesResponse` - Response message for ListBaselineAggregates Up

baselineAggregates (optional)

array[v2BaselineAggregation]

nextPageToken (optional)

String Used to retrieve the next page of results. When unset, no further items are available (this response was the last page).

`v2ListDeploymentsResponse` - Response message for ListDeployments Up

deployments (optional)

array[v2Deployment] Collection of deployments.

nextPageToken (optional)

String Used to retrieve the next page of results. When unset, no further items are available (this response was the last page).

`v2ListEndpointsResponse` - Up

The response message for listing endpoints.

endpoints (optional)

array[v2Endpoint] The endpoints.

nextPageToken (optional)

String If present, signals that more results from the request are available. Use this as another request's page_token value. The field values of subsequent requests must be identical to the previous one.

`v2ListExperimentArtifactCompositionsResponse` - The response message to return list of artifact compositions. (-- api-linter: core::0158::response-next-page-token-field=disabled --) Up

experimentArtifactCompositions (optional)

array[v2ArtifactComposition] List of artifact composition entities.

`v2ListSecurityOptionTypesResponse` - (-- api-linter: core::0158::response-next-page-token-field=disabled aip.dev/not-precedent: We need to do this because return always all options. --) Response method for ListSecurityOptionTypes Up

securityOptionTypes (optional)

array[v2SecurityOptionType] List of security option types activated for the environment.

`v2LogicalType` - Up

Logical types.

LOGICAL_TYPE_UNSPECIFIED: Unspecified type
NUMERICAL: Numerical type
CATEGORICAL: Categorial type
DATETIME: Datetime type
TEXT: Text type
IMAGE: Image type
TIMESTAMP: Timestamp type
ID: ID type

`v2MaxUnavailable` - Represents maximum unavailability configuration Up

pods (optional)

Integer format: int32

percentage (optional)

Integer format: int32

`v2MinAvailable` - Represents minimum availability configuration Up

pods (optional)

Integer format: int32

percentage (optional)

Integer format: int32

`v2MissingValues` - Up

Metrics related to missing values.

rowCount (optional)

Integer Number of rows in the dataset that a specific value is missing. format: int32

`v2ModelType` - Up

Represents a known model type.

Models may be packaged differently, or otherwise have different requirements in order to use. A ModelType represents a grouping of entities with the same expectations. (-- api-linter: core::0123::resource-annotation=disabled --)

name (optional)

String Identifying name for the entity.

displayName (optional)

String Human-readable name of the entity.

`v2MonitoringConfig` - Up

Represents a monitoring configuration type.

enabled (optional)

frontendUri (optional)

`v2MonitoringOptions` - Up

MonitoringOptions is the entity which represents a monitoring options for model deployment.

enabled (optional)

columns (optional)

array[v2Column] List of Columns on which aggregations will be performed.

baselineAggregations (optional)

array[v2BaselineAggregation] Baseline aggregation data.

baselineSource (optional)

v2BaselineSource

timestampColumn (optional)

kafkaTopic (optional)

String Kafka topic where raw data will be send. If not present global value will be used.

`v2NumericalAggregate` - Up

Aggregated metrics specific for columns on LogicalType.LOGICAL_TYPE_NUMERICAL type. Aggregated metrics are calculated for a specific time window.

binEdges (optional)

array[Double]

Data points for the window are binned before calculating some metrics. bin_edges will hold the edges of all the bins excluding -inf and +inf. 2 more bins will be formed from the -inf to 1st bin_edge and from the last bin_edge to +inf.

Example: bin_edges - [-3, 0, 3] bins will be - [-inf, -3], [-3, 0], [0, 3], [3, +inf] So always the bin count == len(bin_edges) + 1

format: double

binCount (optional)

array[Integer]

Count of data points fall in to each bin in the order of the bins.

Let's assume bin_edges are [p,q,...,r,s] The 1st count will represent the count of the [-inf, p] bin The last bin will represent the count of the [s, +inf] bin

format: int32

mean (optional)

Double Mean of the data for the aggregated window. format: double

standardDeviation (optional)

Double Standard deviation of the data for the aggregated window. format: double

min (optional)

Double Minimum data point of the aggregated window. format: double

max (optional)

Double Maximum data point of the aggregated window. format: double

sum (optional)

Double Sum of all values in column. format: double

`v2PodDisruptionBudgetSpec` - Up

Represents configuration for Pod Disruption Budget (PDB). Only one of min_available or max_unavailable should be specified.

minAvailable (optional)

v2MinAvailable

maxUnavailable (optional)

v2MaxUnavailable

`v2ResourceUnit` - Up

Enum for supported units in resource specifications.

RESOURCE_UNIT_UNSPECIFIED: Unspecified
CORES: For CPU cores
MILLICORES: For CPU millicores
MIB: For memory in MiB
GIB: For memory in GiB

`v2Runtime` - Represents a runtime capable of processing an entity with a specific `ModelType`. (-- api-linter: core::0123::resource-annotation=disabled --) Up

name (optional)

String Identifying name for the entity.

displayName (optional)

String Human-readable name of the entity.

modelType (optional)

String Name of the ModelType this runtime is compatible with.

image (optional)

String Full name of the container image for the runtime service.

`v2ScoringAggregation` - Aggregated scoring data Up

id (optional)

String Scoring Aggregation ID autogenerated.

column (optional)

deploymentId (optional)

startTime (optional)

Date Start time for aggregation. format: date-time

segmentDuration (optional)

String Aggregation window.

numericalAggregate (optional)

v2NumericalAggregate

categoricalAggregate (optional)

v2CategoricalAggregate

missingValues (optional)

v2MissingValues

isModelOutput (optional)

v2AuthenticationPassphrase

`v2ScoringEndpoint` - Up

ScoringEndpoint defines the HTTP endpoint configuration used to invoke model scoring. It specifies the URL and HTTP method required to send scoring requests to the deployed model. Clients should use this information to construct proper API calls when requesting predictions.

uri (optional)

String POST endpoint URL of the deployment, typically in the form <ip/host>:/. This is the empty string when the endpoint is not available.

method (optional)

v2HttpMethod

`v2ScoringInterface` - Up

Defines available endpoints and API contract of a deployed scoring model.

score (optional)

v2ScoringEndpoint

sampleRequest (optional)

v2ScoringEndpoint

capabilities (optional)

v2ScoringEndpoint

`v2Security` - Up

Represents the security properties of a Deployment, defining authentication mechanisms.

passphrase (optional)

tokenAuth (optional)

v2AuthorizationAccessToken

disabledSecurity (optional)

Object Disables security. This option should be used with caution as it allows unauthenticated access.

updateTime (optional)

Date format: date-time

`v2SecurityOptionType` - Up

Represents a security option type.

disableSecurityType (optional)

Object Disables security.

hashAlgorithm (optional)

AuthenticationPassphraseHashAlgorithm

authorizationProtocol (optional)

AuthorizationAccessTokenAuthorizationProtocolSecurityType

displayName (optional)