README.md

---
license: mit
task_categories:
- text-to-image
language:
- en
size_categories:
- n<1K
configs:
- config_name: default
  data_files:
  - split: test
    path: test.jsonl
---

# Pick-a-Pic v1 — test_unique prompts

The 500 unique prompts from the `test_unique` split of
[Pick-a-Pic v1](https://huggingface.co/datasets/yuvalkirstain/pickapic_v1),
extracted for use as a text-to-image generation prompt set.

This is the split used for the held-out evaluation in the
[PickScore paper](https://arxiv.org/abs/2305.01569)
(*Pick-a-Pic: An Open Dataset of User Preferences for Text-to-Image Generation*).

## Contents

| file | description |
|---|---|
| `test.jsonl` | 500 records: `{"id": int, "prompt": str}` |
| `prompts.txt` | the same 500 prompts, one per line |

Images and human preference labels are **not** included — only the prompts.
For the full data (paired images + preference labels) use the original dataset.

## Usage

```python
from datasets import load_dataset

ds = load_dataset("Jialuo21/pickapic-test", split="test")
print(len(ds))          # 500
print(ds[0]["prompt"])  # 'A curious cat exploring a haunted mansion'
```

## Source

Extracted from the public mirror
[`pickapic-anonymous/pickapic_v1`](https://huggingface.co/datasets/pickapic-anonymous/pickapic_v1)
(`test_unique` split, 500 rows, all prompts unique).

## Citation

```bibtex
@inproceedings{kirstain2023pickapic,
  title     = {Pick-a-Pic: An Open Dataset of User Preferences for Text-to-Image Generation},
  author    = {Kirstain, Yuval and Polyak, Adam and Singer, Uriel and
               Matiana, Shahbuland and Penna, Joe and Levy, Omer},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year      = {2023}
}
```