This dataset is a merged, deduplicated, and balanced structured SFT dataset for Qwen3 4B instruction fine-tuning.
It contains 3350 examples in formats: json, xml, yaml, toml, csv.
License
Add your license here.
Usage
from datasets import load_dataset
ds = load_dataset("Taichi11/structured-sft-boosted_v2")
