Dataset Tools

Overview

In Arize, datasets are collections of examples used for experimentation and evaluation. Each dataset contains rows of data organized by columns, with support for string, numeric, and list-of-string values.

In arize_toolkit, the Client exposes helpers for:

1. Listing all datasets in a space
2. Retrieving dataset metadata by name or ID
3. Fetching all example rows from the latest dataset version

Operation	Helper
List all datasets	get_all_datasets
Get dataset by name	get_dataset
Get dataset by ID	get_dataset_by_id
Get dataset examples	get_dataset_examples

Dataset Operations

---

get_all_datasets

datasets: list[dict] = client.get_all_datasets()

Retrieves all datasets in the current space. Automatically paginates through all results.

Returns

A list of dataset dictionaries, each containing id, name, createdAt, updatedAt, datasetType, status, columns, experimentCount.

Example

datasets = client.get_all_datasets()
for ds in datasets:
    print(f"{ds['name']} ({ds['status']}) - {ds['experimentCount']} experiments")

---

get_dataset

dataset: dict = client.get_dataset(name="my-dataset")

Retrieves a dataset by name within the current space.

Parameters

• name (str) — Name of the dataset to retrieve.

Returns

A dictionary with dataset metadata: id, name, createdAt, updatedAt, datasetType, status, columns, experimentCount.

Example

dataset = client.get_dataset(name="pharmacy-malicious-baseline")
print(dataset["id"])  # "RGF0YXNldDox..."
print(dataset["status"])  # "active"
print(dataset["columns"])  # ["input", "output", "timestamp"]

---

get_dataset_by_id

dataset: dict = client.get_dataset_by_id(dataset_id="RGF0YXNldDox...")

Retrieves a dataset by its ID.

Parameters

• dataset_id (str) — ID of the dataset (base64-encoded).

Returns

A dictionary with dataset metadata: id, name, createdAt, updatedAt, datasetType, status, columns, experimentCount.

Example

dataset = client.get_dataset_by_id(dataset_id="RGF0YXNldDox...")
print(dataset["name"])  # "pharmacy-malicious-baseline"

---

get_dataset_examples

examples: list[dict] = client.get_dataset_examples(name="my-dataset")

Retrieves all examples (rows) from the latest version of a dataset. Automatically paginates through all results. Each example contains a data dictionary mapping column names to their values.

Parameters

• name (Optional[str]) — Name of the dataset. Either name or dataset_id must be provided.
• dataset_id (Optional[str]) — ID of the dataset. Either name or dataset_id must be provided.

Returns

A list of example dictionaries, each containing:

• id (str) — Unique identifier for the example
• createdAt (datetime) — When the example was created
• updatedAt (datetime) — When the example was last updated
• data (dict) — Column name to value mapping

Example

examples = client.get_dataset_examples(name="pharmacy-malicious-baseline")
for ex in examples:
    print(ex["id"])  # "RGF0YXNldEV4YW1wbGU6..."
    print(ex["data"]["input"])  # "What is the dosage for..."
    print(ex["data"]["output"])  # "The recommended dosage..."

# Or by dataset ID
examples = client.get_dataset_examples(dataset_id="RGF0YXNldDox...")
print(len(examples))  # 50