L3Cube-MahaParaphrase

L3Cube-MahaParaphrase

MH Specific

L3Cube-MahaParaphrase dataset for language nlp.

Build a semantic search engine for Marathi educational content that finds similar questions and answers across textbooks.
HomepageGitHub

Quick Start

from datasets import load_dataset
ds = load_dataset('l3cube-pune/MahaParaphrase')
for ex in ds['train'][:5]:
    print(f"S1: {ex['sentence1'][:60]}...")
    print(f"S2: {ex['sentence2'][:60]}...")
    print(f"Paraphrase: {bool(ex['label'])}\n")
Modality
text
Size
8,000 sentence pairs
License
Format
CSV
Language
mr
Update Frequency
static
Organization
L3Cube, Pune
Category
language-nlp

Schema

FieldTypeDescription
sentence1stringFirst Marathi sentence
sentence2stringSecond Marathi sentence
labelintWhether the sentences are paraphrases (1) or not (0)

Build With This

Create a question deduplication system for Marathi Q&A platforms that merges duplicate questions and consolidates answers
Develop a FAQ matching service that instantly finds whether a customer's Marathi query already has a documented answer
Build a plagiarism detection tool for Marathi academic submissions that identifies paraphrased content

AI Use Cases

Paraphrase detection
Last verified: 2026-03-07