Dataset · beyoru
Character-persona reasoning roleplay dialogues
beyoru/Aesir-Character-CoT-roleplay
1,973 English roleplay dialogues with chains of thought (CoT) written in character.
What's inside
The dataset contains 1,973 filtered dialogues in English, including about 14,349 assistant turns. Unlike regular reasoning datasets, the internal monologues (CoT) here are written from the first person of the character itself. The data is distilled using the deepseek-v4-pro model and cleaned of empty answers and prohibited content.
What you can do with it
- SFT fine-tuning of roleplay language models with deep persona retention
- Training game NPCs and chatbots to plan responses within character
- Tasks
- Text generation
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 499
- Likes
- 53
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("beyoru/Aesir-Character-CoT-roleplay")