Skip to content

Dataset · NousResearch

Teacher-student response pairs on values topics

NousResearch/arete-data-pairs

English prompt-response pairs from Inkling and Qwen 3.8-27B models with teacher review for alignment.

What's inside

This dataset contains between 10,000 and 100,000 examples of English prompt-response pairs on seven values topics. For each request, responses were generated by the Inkling teacher model and the starting checkpoint of Qwen 3.8-27B, supplemented with category tags and teacher reviews. The files are provided in jsonl format.

What you can do with it

  • Training alignment models and distilling teacher model behavior
  • Training reward models or DPO on response pairs with quality annotations
Tasks
Text generation, Question answering
Languages
English
Size
10K–100K rows
Downloads this month
2K
Likes
78

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("NousResearch/arete-data-pairs")