By clicking "Agree" I confirm I have read and agree to the PoSTMEDIA Data Agreement for Model Training and agree that I intend to use this data for model training purposes only.

rosetta-ko-math-synth-sft

Korean-native mathematics data — problems with step-by-step Korean solutions, answer-verified by Math-Verify (LaTeX \boxed{} + symbolic equivalence).

Synthetic data generated with the Qwen3.6-27B teacher model — part of the Rosetta-KO suite for the Rosetta Korean LLM (PoSTMEDIA).

Math Suite

Sibling datasets from the same pipeline (each a separate repo):

repo	format
`rosetta-ko-math-synth-sft` (this repo)	supervised fine-tuning (user/assistant `messages`)
`rosetta-ko-math-synth-sft-think`	reasoning SFT (`<think>` chain-of-thought)
`rosetta-ko-math-synth-dpo`	preference pairs (`chosen`/`rejected`)
`rosetta-ko-math-synth-dpo-think`	reasoning preference pairs
`rosetta-ko-math-synth-rlvr`	reinforcement-learning-with-verifiable-reward rows

Data Fields

{"id": "...", "source_dataset": "...", "domain": "math",
 "messages": [{"role": "user", "content": "..."},
              {"role": "assistant", "content": "..."}]}

id — row-level identifier
messages[0] (user) — self-contained Korean prompt
messages[1] (assistant) — the answer directly (no reasoning trace)

Usage

from datasets import load_dataset
ds = load_dataset("PoSTMEDIA/rosetta-ko-math-synth-sft", split="train")

License

Released under the Apache License 2.0.

Seed sources:

Citation

@misc{rosetta_ko_math_synth,
  title  = {Rosetta-KO Math Synthetic Dataset},
  author = {PoSTMEDIA},
  year   = {2026},
  howpublished = {Hugging Face: PoSTMEDIA/rosetta-ko-math-synth-sft}
}

Contact

PoSTMEDIA — https://huggingface.co/PoSTMEDIA

Downloads last month: 35

Total file size:

2.41 GB

Collection including PoSTMEDIA/rosetta-ko-math-synth-sft

Rosetta-Ko Suite

Collection

33 items • Updated 22 days ago