OpenOrca
A 4.2M sample dataset replicating Microsoft Research's Orca paper by augmenting FLAN Collection with GPT-4 and GPT-3.5 explanations. Enables small models to match much larger models through explanation-based fine-tuning.
from datasets import load_dataset
ds = load_dataset("Open-Orca/OpenOrca")Preview a sample row
{
"system_prompt": "You are a helpful, respectful assistant. Think step by step.",
"question": "What causes ocean tides?",
"response": "Ocean tides are primarily caused by the gravitational pull of the Moon and, to a lesser extent, the Sun on Earth's oceans..."
}Fine-tune with this dataset
Estimated VRAM to fine-tune with QLoRA (4-bit base model + LoRA adapters), using conservative defaults:
New to fine-tuning? Follow the step-by-step walkthrough: Fine-Tune Your First LLM in 1 Hour
Frequently asked questions
Can I use OpenOrca commercially?
Yes — OpenOrca is released under MIT, a permissive license that allows commercial use, including training models you ship in a product. Check the dataset card for attribution requirements before release.
How much data does OpenOrca contain, and do I need all of it?
OpenOrca contains 4.2M Samples. You rarely need all of it: for style and format fine-tuning, a few hundred to a few thousand examples are enough — load a slice (e.g. split="train[:1000]") and scale up only if quality plateaus.
What is OpenOrca best used for?
Explanation-style SFT at scale (the Orca recipe). It belongs to the Instruction / SFT section of our dataset hub, where you'll find alternatives and complementary sets.