Manage Data

Guide users through the full dataset lifecycle: create, modify, push, pull, version, and organize benchmark data.

When To Use

• Creating a new benchmark dataset from Python dicts, CSV, or existing data.
• Pushing data to the ZeroEval backend.
• Pulling an existing benchmark dataset for evaluation.
• Working with dataset versions and subsets.
• Setting up git-based data management for a benchmark.
• Adding multimodal data (images, audio, video) to datasets.
• Converting CSV/JSONL to Parquet for git workflows.

Execution Sequence

Follow these steps in order. Load reference files only when the current step requires detailed API usage.

Step 1: Install and Initialize