Stanford Alpaca
The foundational self-instruct dataset that launched a thousand fine-tunes. 52k instruction-following examples generated by GPT-3 (text-davinci-003) using the Self-Instruct technique. Used to train the original Alpaca 7B and inspired nearly every instruction dataset that followed.
from datasets import load_dataset
ds = load_dataset("tatsu-lab/alpaca")Preview a sample row
{
"instruction": "Give three tips for staying healthy.",
"input": "",
"output": "1. Eat a balanced and nutritious diet with plenty of fruits and vegetables.\n2. Exercise regularly — aim for at least 30 minutes of moderate activity most days.\n3. Get enough sleep; most adults need 7–9 hours per night."
}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 Stanford Alpaca commercially?
Not in a product — Stanford Alpaca is released under CC-BY-NC 4.0, which restricts use to research and other non-commercial purposes. For commercial fine-tuning, pick a permissively licensed dataset from the same category instead.
How much data does Stanford Alpaca contain, and do I need all of it?
Stanford Alpaca contains 52k Rows. 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 Stanford Alpaca best used for?
Learning the classic instruction format; research only (CC BY-NC). It belongs to the Instruction / SFT section of our dataset hub, where you'll find alternatives and complementary sets.