Skip to content

Dataset · ise-uiuc

Decontaminated 110K code instruction dataset

ise-uiuc/Magicoder-Evol-Instruct-110K

A decontaminated version of evol-codealpaca-v1 for training code generation without test leaks.

What's inside

This is a cleaned version of the evol-codealpaca-v1 dataset with approximately 110K instruction examples for code generation. Decontamination was performed in the same way as in StarCoder, via the bigcode process. According to metadata, the size ranges from 100K to 1M examples; the language is not specified in the card.

What you can do with it

  • SFT of a code LLM
  • Training LoRA for code generation
  • Checking for test data leaks
Tasks
Text generation
Size
100K–1M rows
Downloads this month
25.8K
Likes
198

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("ise-uiuc/Magicoder-Evol-Instruct-110K")