Skip to content

Dataset · databricks

Databricks Dolly 15k instructions

databricks/databricks-dolly-15k

15k English instruction-following records across InstructGPT categories for model fine-tuning and evaluation.

What's inside

An open set of instruction-following records created by thousands of Databricks employees. Categories are taken from the InstructGPT paper: brainstorming, classification, closed QA, generation, information extraction, open QA, and summarization. Licensed under CC BY-SA 3.0 in English.

What you can do with it

  • SFT for instruction-following
  • LoRA training
  • Evaluating models on QA and summarization
Tasks
Question answering, Summarization
Languages
English
Size
10K–100K rows
Downloads this month
62K
Likes
1.4K

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("databricks/databricks-dolly-15k")