MWirelabs/khasi-english-semantic-search

Khasi-English Semantic Search Model

First production-ready semantic search model for Khasi-English language pairs.

Overview

This model enables semantic search between English and Khasi languages, supporting Northeast India's linguistic diversity. Trained on 66,794 English-Khasi translation pairs.

Use Cases

• Cross-lingual semantic search (English ↔ Khasi)
• Document similarity in bilingual contexts
• Cultural content discovery for Northeast India
• Educational language learning tools

Performance

• English-Khasi similarity: 0.69-0.74
• Model size: ~90MB (lightweight deployment)
• 384-dimensional embeddings

Quick Start

from sentence_transformers import SentenceTransformer

model = SentenceTransformer('MWirelabs/khasi-english-semantic-search')
sentences = ['Hello', 'hangne', 'Good morning']
embeddings = model.encode(sentences)

Developed by MWirelabs for Northeast India AI innovation.