gimmy256/math-code-africa

Math & Code — African Context Dataset

Instruction-tuning dataset covering math and coding in African contexts: word problems with African currencies (UGX, KES, NGN), names, geography, and real-world scenarios (mobile money, market trading, farming); coding challenges for USSD systems, mobile money APIs, SMS gateways, agricultural data pipelines, and multilingual NLP — grounded via web search, generated with gemini-2.5-flash.

Dataset Details

• —Rows: 331
• —Regions covered: Uganda, Kenya, Tanzania, Rwanda, Nigeria, Ghana, Ethiopia, South Africa, pan-African
• —Generation model: gemini-2.5-flash (Google Search grounding + document understanding)
• —Format: Parquet, instruction-tuning schema
• —Fields: instruction, input, output, region, topic, category, source_title, source_url, seed_query, generated_by, generated_at

Sourcing & Grounding

Records were generated via web-grounded search queries and/or local reference documents. Each record retains its source title/URL where available for traceability and downstream verification. Records without a verifiable grounded source were discarded rather than fabricated.

Intended Use

Fine-tuning and evaluation of language models on African-context reasoning for the Math & Code domain, as part of the Adaption Platform Part 2 Challenge submission (Crane AI Labs / AI Studio Uganda).

Limitations

• —Generated content should be spot-checked before use in high-stakes (e.g. medical, legal, financial) downstream applications.
• —Regional coverage is not perfectly balanced across all listed countries; see the region column distribution before use.

Citation

Generated by Crane AI Labs / AI Studio Uganda for the Adaption Platform Part 2 Challenge, August 2026.