AI for education
Luganda-Linguistic-Knowledge-Benchmark
Luganda Linguistic Knowledge (LLK) Benchmark
Tests whether the model actually knows the Luganda language rules it is supposed to teach. Structured around CEFR levels with 75% of questions at foundational levels (A1–B1), heavily weighted toward Morphology & Concord (30%) and Syntax (25%) given Luganda's 12-noun-class agreement system. Includes C1–C2 stress tests for cultural context and advanced grammar. 100 mixed questions per language: multiple-choice (51), short-form (47), and… See the full description on the dataset page: https://huggingface.co/datasets/AI-for-Education/Luganda-Linguistic-Knowledge-Benchmark.pedagogy-benchmark
Dataset Card for The Pedagogy Benchmark
Dataset Summary
This dataset provides the questions for the pedgagoy benchmarks described in Benchmarking the Pedagogical Knowledge of Large Language Models.
These are questions from teacher training exams, which are designed to evaluate large language models on their Cross-Domain Pedagogical Knowledge (CDPK) and Special Education Needs and Disability (SEND) pedagogical knowledge.
Existing benchmarks have largely focused on content… See the full description on the dataset page: https://huggingface.co/datasets/AI-for-Education/pedagogy-benchmark.Luganda-Linguistic-Pedagogical-Knowledge-Benchmark
Luganda Linguistic Pedagogical Knowledge (LLPK) Benchmark
Multiple-choice benchmark measuring whether a model understands how to teach foundational literacy in the Ugandan context. Built via a hybrid LLM-generation + human-review pipeline grounded in the global reading science (incl. the GEEAP report), the Ministry of Education P1 Luganda Teacher's Guide, and Pilkington's 1915 A Handbook of Luganda. Generated with Gemini 2.5 Flash and validated by an education expert and native… See the full description on the dataset page: https://huggingface.co/datasets/AI-for-Education/Luganda-Linguistic-Pedagogical-Knowledge-Benchmark.English-Linguistic-Knowledge-Benchmark
English Linguistic Knowledge Benchmark
A 20-item public sample of the English Linguistic Knowledge Benchmark,
measuring how much an LLM knows about the linguistics of English (phonics/
orthography, morphology, syntax, vocabulary, and phonological awareness),
rather than its ability to converse in the language.
This is a curated sample, not the full item set: the full 137-item benchmark
(99 multiple_choice, 38 short_form)
is kept private to avoid leaking into future models'… See the full description on the dataset page: https://huggingface.co/datasets/AI-for-Education/English-Linguistic-Knowledge-Benchmark.Kiswahili-Linguistic-Knowledge-Benchmark
Kiswahili Linguistic Knowledge Benchmark
A 20-item public sample of the Kiswahili Linguistic Knowledge Benchmark,
measuring how much an LLM knows about the linguistics of Kiswahili (phonics/
orthography, morphology, syntax, vocabulary, and phonological awareness),
rather than its ability to converse in the language. Questions are written in English; only the linguistic subject of the questions is Kiswahili.
This is a curated sample, not the full item set: the full 135-item… See the full description on the dataset page: https://huggingface.co/datasets/AI-for-Education/Kiswahili-Linguistic-Knowledge-Benchmark.