Skip to main content
GET
Get alignment task

Authorizations

Authorization
string
header
required

Most Arize AI endpoints require authentication. For those endpoints that require authentication, include your API key in the request header using the format

Path Parameters

alignment_task_id
string
required

The unique alignment task identifier (base64) A universally unique identifier (base64-encoded opaque string).

Example:

"RW50aXR5OjEyMzQ1"

Response

Returns the new alignment task

A reusable comparison of evaluator labels and human annotations on a dataset. The dataset, label column, annotation column, and comparison function identify the task. The name does not affect task identity.

id
string
required

The unique identifier for the alignment task

created_at
string<date-time>
required

When the alignment task was created

updated_at
string<date-time>
required

When the alignment task was last modified

name
string
required

The name of the alignment task

dataset_id
string
required

The identifier of the dataset that contains the labels and annotations

judge_label_column
string
required

The dataset column holding the evaluator's labels, for example eval.hallucination.label when the labels are already on the dataset, or output.label when an experiment produces them.

annotation_column
string
required

The dataset column holding the human annotations to compare against

alignment_function
enum<string>
required

How an alignment task compares a judge label to a human annotation.

  • EXACT_MATCH - Labels agree when their values are equal. The score is the fraction of comparable rows that agree.
Available options:
EXACT_MATCH