jlm429/bluebird-house-synthetic-dataset

# Bluebird House Synthetic Dataset

## Description
This is a small synthetic dataset containing structured facts about a fictional "Bluebird House."

Each example consists of:
- an instruction (question)
- a structured response (Location / Details format)

The dataset is designed to demonstrate:
- LoRA fine-tuning
- structured output learning
- controlled memorization in language models

---

## Format

```json
{
"instruction": "Where is the main water shutoff in the Bluebird House?",
"output": "Location: basement utility room\nDetails: north wall, next to the red pressure tank"
}
```

---

## Use Case

This dataset is intended for:
- instructional fine-tuning experiments
- demonstrating LoRA / PEFT workflows
- testing model behavior on synthetic structured knowledge

---

## Notes

- All data is synthetic and fictional
- Designed to create a clear before/after fine-tuning signal