Skip to content

Dataset · microsoft

Synthetic instructions for code editing

microsoft/NextCoderDataset

Around 381 thousand English examples for training code-editing models across eight programming languages.

What's inside

It contains a synthetic instruction dataset of approximately 381 thousand examples (127k*3) in English. The data covers eight programming languages: Python, Java, C++, C, Rust, Javascript, Go, and Kotlin. The dataset was created by Microsoft authors for SFT training of the NextCoder model family.

What you can do with it

  • Fine-tuning (SFT) language models for code editing and refactoring tasks
  • Creating LoRA adapters for multilingual AI developer assistants
Tasks
Text generation
Languages
English
Size
100K–1M rows
Downloads this month
721
Likes
56

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("microsoft/NextCoderDataset")