Datasets:

guicybercode
/

china-brazil-history-corpus

language:
  - pt
  - en
  - zh
license: apache-2.0
tags:
  - history
  - china
  - brazil
  - small-llm
  - pretraining
  - fine-tuning
  - edge-ai
size_categories:
  - n&lt;1K

China-Brazil History Corpus

A curated text corpus for training and fine-tuning small language models on Chinese and Brazilian history.

What's inside

15 entries (MVP version, expanding to 5,000)
Categories: Dynasties, colonization, immigration, revolutions, contemporary history, philosophy, inventions
Regions: China, Brazil, both
Types: Narratives, timelines, biographies, event analyses, QA pairs
Languages: Portuguese (60%), English (30%), Chinese (10%)

Topics covered

China	Brazil
Xia, Shang, Zhou, Qin, Han dynasties	Pre-Cabraline period & indigenous peoples
Three Kingdoms, Tang, Song, Yuan	Portuguese colonization (1500-1822)
Ming, Qing dynasties	Empire & First Republic
Xinhai Revolution, Republic	Vargas Era, Military Dictatorship
Mao era, Cultural Revolution	New Republic (1985-present)
Contemporary China (1978-present)	Economic & social history
Philosophy (Confucianism, Taoism, Buddhism)	Social movements
Inventions & discoveries	Sino-Brazilian relations

Use cases

Continual pre-training of small LLMs (≤1B params)
Fine-tuning for history QA and summarization
Edge deployment on Mac M4 / 16GB RAM

Format

JSONL with metadata fields:

text: Historical content
category: Thematic category
region: china | brazil | both
period: Historical period or year
source_type: narrative | timeline | biography | event_analysis | qa_pair | comparison

Citation

If you use this dataset, please cite:

@dataset{guicybercode_china_brazil_history,
  author = {Guilherme Monteiro},
  title = {China-Brazil History Corpus},
  year = {2026},
  publisher = {Hugging Face}
}

License
Apache 2.0