Skip to content

Dataset · grammarly

CoEdIT: instruction-based text editing

grammarly/coedit

A dataset for training models on instruction-based text editing in English.

What's inside

The dataset was used to train CoEdIT models for text editing using task-specific instruction tuning. It contains texts and instructions for editing tasks; details are in the CoEdIT paper. The language is English, and the size is 10K–100K.

What you can do with it

  • Fine-tune LLMs for text editing
  • LoRA for instruction tuning
  • Evaluating editing quality
Tasks
Text generation
Languages
English
Size
10K–100K rows
Downloads this month
2.1K
Likes
101

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("grammarly/coedit")