nassimjp/riddles-multilingual

🧩 Riddles Multilingual 1K

A curated multilingual riddle dataset containing 1,000 riddles per language across 5 languages — 5,000 riddle questions in total.

Designed for:

• 🧠 Language learning & brain training
• 🤖 LLM fine-tuning (low-resource + high-resource languages)
• 📚 Educational apps & kids' quizzes
• 🎯 Trivia, riddle, and puzzle generation
• 🌍 Cross-lingual research (Pashto, Urdu, Farsi/Dari, English, Japanese)

---

📦 Dataset Structure

riddles-multilingual/
└── data/
    ├── riddles_pashto_1k.jsonl          (1000 lines)
    ├── riddles_urdu_1k.jsonl            (1000 lines)
    ├── riddles_farsi_dari_1k.jsonl      (1000 lines)
    ├── riddles_english_1k.jsonl         (1000 lines)
    └── riddles_japanese_1k.jsonl        (1000 lines)

---

🗣️ Languages Covered

Config	File	Language	Script	Rows
pashto	riddles_pashto_1k.jsonl	Pashto	Arabic (Naskh)	1000
urdu	riddles_urdu_1k.jsonl	Urdu	Arabic (Nastaliq)	1000
farsi_dari	riddles_farsi_dari_1k.jsonl	Farsi / Dari	Arabic (Naskh)	1000
english	riddles_english_1k.jsonl	English	Latin	1000
japanese	riddles_japanese_1k.jsonl	Japanese	Kana + Kanji	1000

---

🧾 Data Format

Each file is JSONL — one JSON object per line:

{"id": 1, "question": "زه په ورځ کې روښانه یم خو د لمر څخه نه یم؟"}
{"id": 2, "question": "هغه څه شی دی چې په کور کې اوسیږي خو له کوره بهر ګرځي؟"}

Fields

Field	Type	Description
id	int	Riddle number (1–1000)
question	string	Riddle question in the target language

⚠️ This version contains questions only. Answer keys are planned for a future release.

---

🚀 Quick Load (🤗 datasets)

from datasets import load_dataset

# Load one language
ds = load_dataset("nassimjp/riddles-multilingual", "pashto", split="train")
print(ds[0])
# {'id': 1, 'question': 'زه په ورځ کې روښانه یم خو د لمر څخه نه یم؟'}

# Load all languages
ds = load_dataset("nassimjp/riddles-multilingual")
print(ds)

---

📊 Dataset at a Glance

Metric	Value
Total riddles (per language)	1,000
Total languages	5
Total riddle entries	5,000
Format	JSONL
License	CC-BY-4.0
Duplicates	None (validated by ID)
ID range	1 – 1000

---

🎯 Intended Uses

• ✅ Language learning apps (Pashto, Urdu, Farsi, English, Japanese)
• ✅ Kids' educational content — safe, non-violent riddles
• ✅ LLM instruction tuning — multilingual QA pairs
• ✅ Low-resource NLP research — Pashto & Dari are underrepresented
• ✅ Trivia & quiz generators
• ✅ Cross-lingual riddle alignment (parallel IDs 1–1000)

---

⚠️ Limitations

• Riddles are questions only (answers not included yet)
• Some riddles rely on cultural knowledge (Afghan, Islamic, South Asian references)
• Japanese set uses なぞなぞ (nazo-nazo) style — informal
• Not all riddles translate 1:1 across languages due to wordplay
• Coverage is general/educational, not domain-specific

---

🗂️ Riddle Categories (Sample)

• Nature & weather
• Animals, birds, insects
• Human body & senses
• Emotions & abstract concepts
• Food & kitchen
• Household objects
• Technology & AI
• Space & planets
• Islamic & cultural knowledge
• Geography (Afghanistan-focused)
• Sports, tools, professions
• Classic paradox riddles

---

🛣️ Roadmap

• v1.0 — 1,000 riddles × 5 languages (questions only)
• v1.1 — Add answer keys (all languages)
• v1.2 — Add category & difficulty metadata
• v1.3 — Add Dari (Afghan Persian) as separate config
• v1.4 — Expand to 5,000 riddles per language
• v2.0 — Add audio (TTS) for language learning

---

📜 License

Released under CC-BY-4.0 — free to use, share, and adapt with attribution.

---

🙏 Citation

If you use this dataset, please cite:

@misc{riddles_multilingual_1k,
  title        = {Riddles Multilingual 1K},
  author       = {Nassim},
  year         = {2026},
  howpublished = {\url{https://huggingface.co/datasets/nassimjp/riddles-multilingual}},
  note         = {1,000 riddles × 5 languages (Pashto, Urdu, Farsi/Dari, English, Japanese)}
}

---

🤝 Contributing

Contributions welcome:

• 🐛 Fix translation errors
• ➕ Add new languages (Dari, Arabic, Turkish, Hindi)
• 🧠 Add answer keys
• 🏷️ Add category/difficulty tags

Open a PR or issue on the dataset page.

---

📬 Contact

Repo: nassimjp/riddles-multilingual

Made with ❤️ for multilingual learning and low-resource language AI.

---

## ✅ How to Use

1. Go to:  
   👉 https://huggingface.co/datasets/nassimjp/riddles-multilingual/new/main?filename=README.md

2. **Paste the whole block above** (from `---` to the end).

3. Click **Commit new file**.

4. Done ✅ — your dataset card will render with tags, configs, and a proper loading snippet.

---

## 🏷️ Tags Included (for HF discoverability)

`riddles`, `riddle`, `multilingual`, `trivia`, `brain-teasers`, `puzzle`, `pashto`, `urdu`, `farsi`, `dari`, `persian`, `english`, `japanese`, `question-answering`, `llm-training`, `instruction-tuning`, `low-resource-languages`, `afghanistan`, `education`, `kids`, `quiz`, `nazo-nazo`

---