mteb/

IndicQARetrieval

IndicQARetrieval An MTEB dataset Massive Text Embedding Benchmark

IndicQA is a manually curated cloze-style reading comprehension dataset that can be used for evaluating question-answering models in 11 Indic languages. It is repurposed retrieving relevant context for each question.

t2t

Domains Web, Written

Reference https://arxiv.org/abs/2212.05409

You can evaluate an embedding model on this dataset using the… See the full description on the dataset page: https://huggingface.co/datasets/mteb/IndicQARetrieval.