Tag datasets and experiments, edit examples, and manage annotation queues.
For task-oriented walkthroughs of these operations, see the Dataset and annotation guide . Every mutation below is sent as a POST to https://app.arize.com/graphql with an x-api-key header; see Forming calls .
Mutations in this page Reference addTagsToDataset(input: AddTagsToDatasetInput!): AddTagsToDatasetPayloadArguments input AddTagsToDatasetInput!
required
Show AddTagsToDatasetInput fields
Returns Union of TagAssociationSuccess, TagAssociationError. Use inline fragments (... on TypeName).
Example Only required input fields are shown. Replace <ID> and <string> placeholders with real values; the object graph page shows how to look IDs up.
removeTagsFromDataset(input: RemoveTagsFromDatasetInput!): RemoveTagsFromDatasetPayloadArguments input RemoveTagsFromDatasetInput!
required
Show RemoveTagsFromDatasetInput fields
Returns result RemoveTagsFromDatasetResult!
Union of TagAssociationSuccess, TagAssociationError. Use inline fragments (... on TypeName).
Example Only required input fields are shown. Replace <ID> and <string> placeholders with real values; the object graph page shows how to look IDs up.
addTagsToExperiment(input: AddTagsToExperimentInput!): AddTagsToExperimentPayloadArguments input AddTagsToExperimentInput!
required
Show AddTagsToExperimentInput fields
Returns result AddTagsToExperimentResult!
Union of TagAssociationSuccess, TagAssociationError. Use inline fragments (... on TypeName).
Example Only required input fields are shown. Replace <ID> and <string> placeholders with real values; the object graph page shows how to look IDs up.
removeTagsFromExperiment(input: RemoveTagsFromExperimentInput!): RemoveTagsFromExperimentPayloadArguments input RemoveTagsFromExperimentInput!
required
Show RemoveTagsFromExperimentInput fields
Returns result RemoveTagsFromExperimentResult!
Union of TagAssociationSuccess, TagAssociationError. Use inline fragments (... on TypeName).
Example Only required input fields are shown. Replace <ID> and <string> placeholders with real values; the object graph page shows how to look IDs up.
updateDatasetVersionExamples Update the examples of a dataset version
updateDatasetVersionExamples(input: UpdateDatasetVersionExamplesInput!): UpdateDatasetVersionExamplesPayloadArguments input UpdateDatasetVersionExamplesInput!
required
Show UpdateDatasetVersionExamplesInput fields
The id of the dataset version to update
The updates to apply to the examples Show ExampleUpdate fields
The patches to apply to the example The name of the column to patch
dataType DimensionDataType!
required
The data type of the column to patch, used to parse the value One of: STRING, LONG, FLOAT, DOUBLE, EMBEDDING, STRING_LIST, DICTIONARY.
The stringified updated value, the data type is used to parse it
The new examples to add to the dataset version, copied from a previous model or project. Each set selects records or whole sessions. Show NewExampleSet fields
The id of the model or project the examples came from
modelEnvironment ModelEnvironmentName!
required
The environment where the examples came from One of: production, validation, training, tracing.
The start date of the interval where the examples exist
The end date of the interval where the examples exist
Filters to apply to the interval where the examples exist to narrow down the selection (e.g., by model version or batch id) Show FilterItemInput fields
The type of filter. One of: featureLabel, tagLabel, predictionValue, actuals, modelVersion, batchId, predictionClass, predictionScore, actualClass, actualScore, topKPercentile, spanProperty, llmEval, annotation, userAnnotation.
operator ComparisonOperator!
required
The operator of the filter. One of: greaterThan, lessThan, equals, notEquals, greaterThanOrEqual, lessThanOrEqual, topN, contains, containsString, similarTo.
The model dimension to filter, e.g. Age. Nested DimensionInput (same shape as above).
The dimension values of the filter. Nested DimensionValueInput (same shape as above).
The binary values of the prediction.
The numeric values of the prediction.
The categorical values of the prediction.
Filters to apply to the dataset. Compatible with existing filters, joined with AND.
Copy individual records (spans/predictions) by id. Mutually exclusive with sessions and the legacy recordIds. Show RecordSelectionInput fields
predictionID for inference records or span id for tracing records. An empty list selects every record matching the set’s interval and filters.
Copy whole sessions; every span of each selected session becomes an example. Mutually exclusive with records and the legacy recordIds. Show SessionSelectionInput fields
The ids of the sessions to copy. An empty list selects every session the Sessions page shows for the set’s interval and filters.
Superseded by records. The original ids of the records in the project or model to add to the dataset version. Deprecated: Superseded by the records field
copyFromDatasetVersions [CopyExamplesFromDatasetVersionInput!]
New examples to add to the dataset version, copied by id from one or more source dataset versions entirely server-side (no display truncation) Show CopyExamplesFromDatasetVersionInput fields
The id of the dataset version to copy the examples from
The ids of the examples to copy from the source
The ids of the examples to delete from the dataset version
Returns result UpdateDatasetVersionExamplesResponse!
The updated dataset version (with any sampling metadata) or an error Union of UpdateDatasetVersionExamplesError, DatasetVersionWithSampling. Use inline fragments (... on TypeName).
Example Only required input fields are shown. Replace <ID> and <string> placeholders with real values; the object graph page shows how to look IDs up.
createAnnotationConfig Create an annotation config
createAnnotationConfig(input: CreateAnnotationConfigInput!): CreateAnnotationConfigPayloadArguments input CreateAnnotationConfigInput!
required
Show CreateAnnotationConfigInput fields
The id of the space to create the annotation config in
The name of the annotation config
annotationConfigArgs AnnotationConfigArgs!
required
Show AnnotationConfigArgs fields
annotationConfigType AnnotationConfigType!
required
The type of the annotation config One of: categorical, continuous, freeform.
The configuration for a categorical annotation config Show CategoricalConfigInput fields
labelOptions [LabelOptionInput!]!
required
A list of label options Nested LabelOptionInput (same shape as above).
optimizationDirection OptimizationDirection!
required
The optimization direction for the annotation config One of: minimize, maximize, none.
The configuration for a continuous annotation config Show ContinuousConfigInput fields
optimizationDirection OptimizationDirection!
required
The optimization direction for the annotation config One of: minimize, maximize, none.
Returns The created annotation config or error Union of AnnotationConfig, CreateAnnotationConfigError. Use inline fragments (... on TypeName).
Example Only required input fields are shown. Replace <ID> and <string> placeholders with real values; the object graph page shows how to look IDs up.
createAnnotationQueue Create an annotation queue
createAnnotationQueue(input: CreateAnnotationQueueInput!): CreateAnnotationQueuePayloadArguments input CreateAnnotationQueueInput!
required
Show CreateAnnotationQueueInput fields
The id of the space to create the queue in.
[DEPRECATED] The id of the dataset version from which examples will be added to the queue. If not provided, the queue will be empty. Deprecated: Use recordSpecifications for more flexible record selection.
New annotation configs to create and associate with the queue. Show AnnotationConfigInput fields
The name of the annotation config.
annotationConfigType AnnotationConfigType!
required
The type of the annotation config One of: categorical, continuous, freeform.
The configuration for a categorical annotation config Show CategoricalConfigInput fields
labelOptions [LabelOptionInput!]!
required
A list of label options Nested LabelOptionInput (same shape as above).
optimizationDirection OptimizationDirection!
required
The optimization direction for the annotation config One of: minimize, maximize, none.
The configuration for a continuous annotation config Show ContinuousConfigInput fields
optimizationDirection OptimizationDirection!
required
The optimization direction for the annotation config One of: minimize, maximize, none.
The ids of existing annotation configs to associate with the queue.
annotatorAssignment AnnotatorAssignment!
required
The annotators assigned to the queue. Either a specific list of user ids or all annotators in the space. Show AnnotatorAssignment fields
The ids of the users in the space assigned to annotate the queue.
assignmentMethod AnnotationQueueAssignmentMethod!
required
The method used to assign records in the queue to annotators. One of: all, random.
[DEPRECATED] The ids of the records to create the queue with. If provided, the queue will be populated with the corresponding records from the dataset version. Deprecated: Use recordSpecifications for more flexible record selection.
The instructions for the queue.
recordSpecifications [AnnotationQueueRecordSpecificationInput!]
The specifications for the records to be added to the queue. Show AnnotationQueueRecordSpecificationInput fields
sourceType RecordSourceType!
required
The source type of the records to add to the queue. One of: spans, dataset.
The ids of the records to add to the annotation queue. If no records are provided, then all records that match the other criteria will be added.
The start date of the interval where the records exist. Required if the source type is spans.
The end date of the interval where the records exist. Required if the source type is spans.
The dataset version id. Required if the source type is dataset.
The model id. Required if the source type is spans.
The maximum number of records that can be in the queue. Must be between 1 and 5000.
The record column names annotators are allowed to see in the queue. If omitted or empty, annotators see all columns.
Returns returns the created annotation queue or error Union of AnnotationQueue, AnnotationQueueError. Use inline fragments (... on TypeName).
Example Only required input fields are shown. Replace <ID> and <string> placeholders with real values; the object graph page shows how to look IDs up.
updateAnnotations Update the annotations of a dataset
updateAnnotations(input: UpdateAnnotationsInput!): UpdateAnnotationsPayloadArguments input UpdateAnnotationsInput!
required
Show UpdateAnnotationsInput fields
annotationUpdates [AnnotationWithConfigIdInput!]!
required
The annotations to update Show AnnotationWithConfigIdInput fields
The ID of the annotation config
The annotation to update Show AnnotationInput fields
The name of the annotation
Who is updating the annotation
The label of the annotation
The score of the annotation
The freeform text of the annotation
The type of the annotation One of: Label, Score, Text.
[DEPRECATED] The note to update the annotation with. Deprecated: Use freeform text annotations instead. The model record context. If provided, the annotation updates will be written on the record. Show ModelRecordContextInput fields
The id of the model to update the annotations on.
modelEnvironment ModelEnvironmentName!
required
One of: production, validation, training, tracing.
The record id of the span/trace to update the annotations on.
Whether the record is a span or a trace. One of: span, trace, session.
experimentRunContext ExperimentRunContextInput
The experiment run context. If provided, the annotation updates will be written on the experiment run. Show ExperimentRunContextInput fields
The id of the experiment run
datasetRecordContext DatasetRecordContextInput
The dataset record context. If provided, the annotation updates will be written to the dataset record. Show DatasetRecordContextInput fields
The id of the dataset version to update the annotations on.
The record id of the dataset record to update the annotations on
Returns result UpdateAnnotationResultType!
The updated result Union of UpdateAnnotationSuccess, UpdateAnnotationError. Use inline fragments (... on TypeName).
Example Only required input fields are shown. Replace <ID> and <string> placeholders with real values; the object graph page shows how to look IDs up.
batchUpdateAnnotations Batch update the annotations of a dataset.
batchUpdateAnnotations(input: BatchUpdateAnnotationsInput!): BatchUpdateAnnotationsPayloadArguments input BatchUpdateAnnotationsInput!
required
Show BatchUpdateAnnotationsInput fields
The id of the model to update the annotations on.
recordAnnotationUpdates [RecordAnnotationUpdateInput!]!
required
The records to update with their annotations. Show RecordAnnotationUpdateInput fields
The record id of the span/trace to update the annotations on.
The timestamp of the record.
annotationUpdates [AnnotationWithConfigIdInput!]!
required
The annotation updates for this record. Show AnnotationWithConfigIdInput fields
The ID of the annotation config
The annotation to update Nested AnnotationInput (same shape as above).
[DEPRECATED] Optional note to add to this record. Deprecated: Use freeform text annotations instead. Whether the batch contains spans or traces. One of: span, trace, session.
Returns result BatchUpdateAnnotationsResultType!
The update result Union of BatchUpdateAnnotationSuccess, BatchUpdateAnnotationError. Use inline fragments (... on TypeName).
Example Only required input fields are shown. Replace <ID> and <string> placeholders with real values; the object graph page shows how to look IDs up.
