Skip to main content

​
Create from Python data

Use ze.Dataset(name, data=[...]) when your rows already exist in memory. 
import zeroeval as ze

ze.init()

dataset = ze.Dataset(
    "capital-cities",
    data=[
        {"row_id": "fr", "question": "Capital of France?", "answer": "Paris"},
        {"row_id": "de", "question": "Capital of Germany?", "answer": "Berlin"},
    ],
    description="Simple geography dataset",
)
Add a stable row_id field whenever possible. It makes resume behavior and row-level tracking much more reliable.

​
Create from CSV

You can initialize directly from a CSV path. The dataset name defaults to the CSV filename stem. 
import zeroeval as ze

ze.init()

dataset = ze.Dataset("/path/to/questions.csv")
print(dataset.name)  # e.g. "questions"
print(len(dataset))

​
Modify rows before running evals

The Dataset object supports in-memory row operations. 
# Add rows
dataset.add_rows(
    [{"row_id": "es", "question": "Capital of Spain?", "answer": "Madrid"}]
)

# Update one row by index
dataset.update_row(
    0,
    {"question": "Capital city of France?", "answer": "Paris"},
)

# Delete a row by index
dataset.delete_row(1)
Row updates by index can change row ordering and row content. Keep this in mind if your downstream logic assumes fixed order.

​
Push to ZeroEval

Push persists the dataset to the backend and refreshes local rows with backend metadata (including row IDs when available). 
dataset.push()
print(dataset.version_id, dataset.version_number)
Use create_new_version=True if you explicitly want a new version on push: 
dataset.push(create_new_version=True)

​
Validate shape quickly

Useful helpers before running tasks: 
print(dataset.columns)  # sorted list of column names
print(len(dataset))     # number of rows
print(dataset[0])       # DotDict for first row