mteb/

IndicXnliPairClassification

IndicXnliPairClassification An MTEB dataset Massive Text Embedding Benchmark

INDICXNLI is similar to existing XNLI dataset in shape/form, but focusses on Indic language family. The train (392,702), validation (2,490), and evaluation sets (5,010) of English XNLI were translated from English into each of the eleven Indic languages. IndicTrans is a large Transformer-based sequence to sequence model. It is trained on Samanantar dataset (Ramesh et… See the full description on the dataset page: https://huggingface.co/datasets/mteb/IndicXnliPairClassification.