README.md

---
license: cc-by-nc-sa-4.0
language:
- en
- fi
- fr
- de
- sv
- nl
- fa
- da
- es
- gu
- hi
- uk
- ru
- 'no'
- it
configs:
- config_name: test
  data_files:
  - split: test
    path: eloquent-2025-robustness-prompts.json
pretty_name: ELOQUENT Robustness and Consistency Task 2025
size_categories:
- n<1K
---
# ELOQUENT Robustness and Consistency Task
This dataset contains the sample and test datasets for the Robustness and Consistency task, which is part of the ELOQUENT lab. This dataset is for participants to generate texts for prompt variants, to investigate prompt style conditioned variation.

- [Robustness task](https://eloquent-lab.github.io/task-robustness-and-consistency/)
- [ELOQUENT lab](https://eloquent-lab.github.io/)
- [CLEF conference](https://clef2025.clef-initiative.eu/) 9-12 September 2025
  
## The task in brief (this is a simple task to execute!)
- This dataset provides a number of questions in several languages
    - e.g. `"question": "Is it more important to be polite or to be honest?"`
- You use a generative language model to answer the question in the languages your model handles
- Use separate sessions for each response! They are not intended to be interpreted as follow-up responses. 
- You send the response to us before the beginning of June 2025
- We and you together discuss the results to explore how linguistic variation conditions responses
- We write a joint report
- Workshop at CLEF in Madrid 9-12 September 2025

## Submit Here:

[Submission Form](https://forms.gle/cy5hrrWRbyJ8mchz7)

#### Test Data


```python
from datasets import load_dataset
data = load_dataset("Eloquent/Robustness", "test")
```


## Help us translate the test into more languages!

This should be easy and doable in half an hour or so. Just use everyday language. If you use translation aid systems, tell us. Submit the translation here as a pull request or send it directly to the task organisers. 

- Item 013 is nonsense words but the sentence should be reasonably syntactically correct
- Item 014 is all nonsense, but the nonsense should be legible and somewhat pronouncible
- Item 015 mentions Korea, picked more or less at random to be a faraway place. Replace with some other faraway place if Korea is near where your language is spoken!
- Item 018 is intended to probe the worry a foreign visitor in a country might have. It does not need to be literally translated: replace the place name with something else (in Swedish, it is a summer house; in Dutch, a sailing trip) and the activity with something that foreign visitors might need help to navigate.


## Dataset authors

Marie Isabel Engels (en, de)
Jussi Karlgren (en, sv)
Josiane Mothe (fr)
Aarne Talman (fi)
Maria Barrett (da)
Shaghayegh Roohi (fa)
Sander Bijl de Vroe (nl)
Poojan Vachharajani (hi, gu)
Kate Lutsai (uk)
Alba García (es)
Vladislav Mikhailov (ru)
Lilja Øvrelid (no)
Andrea Pasin (it)