Sign in to view source links and access this dataset
Description
116,503 multiple-choice questions curated from various open-source educational websites and textbooks, inspired by the original MMLU dataset. The dataset includes questions from Bangladeshi exams such as job exams, the Bangladesh Civil Service Exam, and undergraduate admission exams. It was created by hishab and last updated on June 20, 2025.
Use Cases
Training language models for multiple-choice question answering based on the described educational content.
Benchmarking model performance on Bangla language comprehension across categories mentioned in the description.
Analyzing question distributions and difficulty across different Bangladeshi exam types described, such as civil service and admission exams.
Strengths
Contains 116,503 questions, providing a substantial collection for model training and evaluation.
Sourced from multiple real-world Bangladeshi exam contexts, including job and civil service exams, likely increasing relevance.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
Curated from various open-source educational websites and textbooks.
Collection Method
Inspired by and following the format of the original MMLU dataset (Hendrycks et al., 2020).
Freshness
Last updated 2025-06-20 17:51:10; freshness should be verified.
Geography
Bangladesh
License is unknown; users should verify terms of use before downloading.