Turkish MCQ Reasoning Dataset for RLVR training
What this dataset is
This dataset is a Turkish multiple-choice (MCQ) reasoning benchmark augmented with a weak-model pass-rate difficulty signal, designed for curriculum learning, GRPO / RLVR-style training, and evaluation of reasoning models on Turkish exam-style questions.
It is built from a Turkish MMLU-style MCQ corpus and annotated with a model-based difficulty proxy (pass_rate), computed via repeated sampling of a weak… See the full description on the dataset page: https://huggingface.co/datasets/barandinho/turkish-mcq-rlvr.
