Easy Dataset
StaleDescription
easy-dataset is a tool to quickly build datasets for LLM fine-tuning and agent training, with auto-labeling, sampling and export to mainstream formats.
Key Features
- One-click generation — input raw text and get QA / SFT datasets
- LLM labeling — high-quality labels via mainstream LLMs
- Multi-format export — Alpaca, ShareGPT, JSONL, Parquet and more
- Visual review — UI for manual QA and edits
- Difficulty stratification — auto-layers by difficulty for curriculum training
- Extensible — plug in custom prompts and filter rules
Use Cases
💡 Quickly producing datasets for domain-specific LLM fine-tuning
💡 Preparing multi-step training samples for agent tasks
💡 Cutting labeling effort with the visual QA UI
💡 Turning existing documents into training datasets in one click
Strengths & Limitations
✅ Strengths
- • High community interest (14.9k stars)
- • Responsive to issues, low backlog
⚠️ Limitations
- • No clear open-source license
Categories
Quick Start
npm install -g @easy-dataset/cli
dataset init my-data
dataset generate --input ./docs --format alpaca