Skip to content

Dataset · HuggingFaceH4

Multilingual chains of thought

HuggingFaceH4/Multilingual-Thinking

1K reasoning examples in Spanish, French, Italian, and German translated from English.

What's inside

A text-generation dataset of reasoning chains translated from English into Spanish, French, Italian, and German. It comprises 1K training examples sampled from the SystemChat subset of SmolTalk2, with reasoning traces translated by another language model. Used in the OpenAI Cookbook for fine-tuning OpenAI gpt-oss models.

What you can do with it

  • Fine-tuning a multilingual reasoning model
  • Training LoRA adapters for gpt-oss on reasoning chains
  • Evaluating cross-lingual CoT
Tasks
Text generation
Languages
English, German, French, Spanish, Italian
Size
1K–10K rows
Downloads this month
7.4K
Likes
121

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("HuggingFaceH4/Multilingual-Thinking")