Skip to content

Dataset · Nopm

Synthetic creative writing data from Claude 3 Opus

Nopm/Opus_WritingStruct

Around 6k cleaned genre texts and writing instruction prompts in English.

What's inside

A compact text dataset (1k to 10k records) in English, generated by the Claude 3 Opus model and cleaned using automated filters. It contains examples of creative prose across a broad spectrum of genres, alongside instruction-following question-answer pairs on writing. Distributed under the permissive Apache 2.0 license.

What you can do with it

  • Supervised fine-tuning (SFT) and LoRA training to improve LLM literary style and roleplay
  • Distilling Claude 3 Opus writing skills into more compact open models
  • Training assistants to support writers and screenwriters
Tasks
Text generation, Question answering
Languages
English
Size
1K–10K rows
Downloads this month
971
Likes
39

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("Nopm/Opus_WritingStruct")