Indian Flag
Government Of India
A-
A
A+
UpvoteUpvote
0

Khasi English Semantic Search Model

Khasi-English semantic search model, trained on 66,794 pairs with 0.69-0.74 similarity. ~90MB, supports Meghalaya tourism/culture. By MWirelabs
  • Upvotes0upvotes
  • Downloads27downloads
  • Views1,196views
  • File size0
  • Shares0Shares
  • Last updatedupdated 01 Sept 2025
UpvoteUpvote
0
LICENSE CC0 1.0 Public Domain
SECTOR Arts, Culture and Tourism

About Model

Developed by MWirelabs, this model is the first production-ready semantic search system for Khasi-English language pairs, celebrating Northeast India’s linguistic diversity, with a special focus on Meghalaya. Trained on a curated corpus of 66,794 English-Khasi translation pairs (63,909 Khasi sentences, 65,239 English sentences, 65,241 parallel pairs), it utilizes the lightweight MiniLM-L6-v2 architecture (~22.7M parameters, ~90MB). The model achieves cosine similarity scores of 0.69-0.74, showcasing effective cross-lingual alignment for Khasi, a low-resource Austroasiatic language spoken primarily in Meghalaya.

The dataset, sourced from cleaned Khasi texts, historical documents, bilingual translations, and cultural/administrative materials from Meghalaya, was preprocessed for anonymization.

Key use cases include cross-lingual document similarity, cultural content discovery (e.g., Meghalaya’s Khasi folklore), and educational tools for the region’s tourism and heritage sectors. The lightweight design supports deployment on low-resource devices, enhancing accessibility in Meghalaya. Ethical considerations emphasize respect for Khasi heritage, encouraging collaboration with Meghalaya’s local communities.

This pioneering effort by MWirelabs, released under Creative Commons CC0 1.0, positions the organization as a leader in Meghalaya and Northeast India’s AI innovation, building on the Khasi-English Word Embeddings model.

Citation: @misc{kajingiathuhsearch2025, title={KaJingïathuhSearch2025: Khasi-English Semantic Search Model}, author={MWirelabs}, year={2025}, publisher={Hugging Face}, howpublished={\url{https://huggingface.co/MWirelabs/khasi-english-semantic-search}} }

Metadata Metadata

CC0 1.0 Public Domain

Arts, Culture and Tourism

MWirelabs

N.A.

MWire Labs

Badal Nyalang

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

Open

18/09/25 10:17:07

N.A.

0

Tags

  • safetensors
  • Sentence Similarity
  • cross-lingual
  • en
  • autotrain_compatible
  • semantic search
  • sentence-transformers
  • license:cc0-1.0
  • kha
  • khasi
  • text-embeddings-inference
  • khasi-culture
  • Meghalaya

Linked Model Linked Model

NE-Embed
NE-Embed is a multilingual text embedding model for Northeast Indian languages, enabling semantic search, retrieval, and RAG across 10 languages including Khasi, Garo, Meitei, Bodo, Mizo, Assamese, Nyishi, Kokborok, Pnar, and Nagamese. Fine-tuned on LaBSE with 201,738 parallel pairs.
MWIRE LABS
  • See Upvoters0
  • Downloads0
  • Views90
  • File Size0


Related Models by Organization Related Models by Organization

Northeast STT Multilingual Speech to Text Model
A multilingual Speech-to-Text (STT) model for eight Northeast Indian languages, fine-tuned from Whisper Medium using over 150,000 speech-text pairs from public and institutional datasets. The model expands speech recognition support for low-resource indigenous languages, including Khasi, Garo, Mizo, Kokborok, Nagamese, Assamese, Chakma, and Wancho.
MWIRE LABS
  • See Upvoters0
  • Downloads4
  • Views163
  • File Size0

NE-SpeechEmbed
NE-SpeechEmbed is a multilingual speech-text embedding model by MWire Labs for Northeast Indian languages. The model supports semantic speech search, cross-modal retrieval, and audio-text embeddings across Khasi, Garo, Mizo, Nagamese, Kokborok, Assamese, Wancho, and Chakma.
MWIRE LABS
  • See Upvoters0
  • Downloads0
  • Views94
  • File Size0

NE-Embed
NE-Embed is a multilingual text embedding model for Northeast Indian languages, enabling semantic search, retrieval, and RAG across 10 languages including Khasi, Garo, Meitei, Bodo, Mizo, Assamese, Nyishi, Kokborok, Pnar, and Nagamese. Fine-tuned on LaBSE with 201,738 parallel pairs.
MWIRE LABS
  • See Upvoters0
  • Downloads0
  • Views90
  • File Size0

Mizo OCR - Text Recognition for Mizo Language
OCR model for the Mizo language achieving 90.68% character accuracy on synthetic and curated printed text
MWIRE LABS
  • See Upvoters0
  • Downloads4
  • Views473
  • File Size0

NE-OCR
NE-OCR is a multilingual Optical Character Recognition model developed by MWire Labs to accurately recognize printed text from documents in Northeast Indian languages. The model supports Assamese, Bodo, English, Garo, Hindi, Khasi, Kokborok, Meitei (Bengali script), Meitei (Meitei Mayek script), Mizo, Nagamese, and Nyishi. It is designed to enable reliable digitization of books, newspapers, government records, educational materials, and cultural archives from Northeast India where mainstream OCR
MWIRE LABS
  • See Upvoters0
  • Downloads9
  • Views320
  • File Size0

Nagamese Speech-to-Text
Automatic Speech Recognition (ASR) model for Nagamese speech, designed to transcribe spoken Nagamese into text for real-world usage.
MWIRE LABS
  • See Upvoters0
  • Downloads5
  • Views301
  • File Size0

Garo OCR - Text Recognition for Garo
OCR model for the Garo language achieving 93.13% character accuracy.
MWIRE LABS
  • See Upvoters0
  • Downloads0
  • Views476
  • File Size0

Northeast Language Identification
NE-LID is a fast and accurate language identification model for Northeast Indian languages using character level features. It is designed for low resource and script diverse text and achieves high accuracy on short sentences.
MWIRE LABS
  • See Upvoters1
  • Downloads13
  • Views1,218
  • File Size0

NortheastNER
NortheastNER is a token classification model built on XLM-RoBERTa and fine-tuned on ~25k sentences from gazetteers, news, and cultural texts across Northeast India. It detects region-specific entities, places, tribes, festivals, tourist sites, flora, fauna, and experimental local names; ideal for low-resource NER, regional search, cultural analytics, and knowledge graph applications.
MWIRE LABS
  • See Upvoters0
  • Downloads15
  • Views497
  • File Size0

Kren-M
Northeast India's first AI language model. Kren-M is a 2.6B parameter bilingual model for Khasi-English, built on Gemma-2-2B. Features Kren-NE custom tokenizer covering 7 NE languages (Khasi, Garo, Mizo, Assamese, Manipuri, Nagamese, Nyishi) with 35.7% efficiency gain. Trained on 5.43M Khasi sentences. Capabilities: bidirectional translation, natural conversation, cultural context. Designed for language preservation across Northeast India
MWIRE LABS
  • See Upvoters0
  • Downloads40
  • Views1,231
  • File Size0


Related Models by Sector Related Models by Sector

NortheastNER
NortheastNER is a token classification model built on XLM-RoBERTa and fine-tuned on ~25k sentences from gazetteers, news, and cultural texts across Northeast India. It detects region-specific entities, places, tribes, festivals, tourist sites, flora, fauna, and experimental local names; ideal for low-resource NER, regional search, cultural analytics, and knowledge graph applications.
MWIRE LABS
  • See Upvoters0
  • Downloads15
  • Views497
  • File Size0

NE-BERT
NE-BERT is Northeast India's first domain-specific multilingual foundation model. Built on the ModernBERT architecture and trained on 8.3 million sentences, it supports 9 regional languages: Assamese, Khasi, Garo, Manipuri (Meitei), Mizo, Nyishi, Nagamese, Kokborok, and Pnar. It achieves State-of-the-Art performance on regional benchmarks and offers 1.6x faster inference, bridging the digital divide for low-resource languages.
MWIRE LABS
  • See Upvoters0
  • Downloads27
  • Views855
  • File Size0


Related Models by Task Type Related Models by Task Type

sarvam-1
India's first indic model, pretrained on 4 trillion tokens
SARVAM AI
  • See Upvoters4
  • Downloads301
  • Views6,806
  • File Size0

AI4Bharat - IndicBERTv2-MLM-Sam-TLM: Multilingual MLM with Sampled Translation
A transformer-based model pre-trained on multiple Indian languages using Masked Language Modeling and Sampling-based Translation Language Modeling techniques.
AI4BHARAT
  • See Upvoters0
  • Downloads6
  • Views269
  • File Size0

AI4Bharat - IndicBERTv2-MLM-only: Multilingual Masked Language Model
A transformer-based model pre-trained exclusively on Masked Language Modeling tasks across multiple Indian languages.
AI4BHARAT
  • See Upvoters0
  • Downloads11
  • Views438
  • File Size0

AI4Bharat - IndicBERT: Multilingual Language Representation Model
A transformer-based model pre-trained on large-scale corpora, covering 12 major Indian languages.
AI4BHARAT
  • See Upvoters1
  • Downloads45
  • Views1,454
  • File Size0

AI4Bharat - IndicBERTv2-SS: Multilingual Sentence Similarity Model
A transformer-based model pre-trained on multiple Indian languages, focusing on Sentence Similarity tasks.
AI4BHARAT
  • See Upvoters0
  • Downloads7
  • Views500
  • File Size0

AI4Bharat - IndicBERTv2-MLM-Back-TLM: Multilingual Language Translation Model with Back-Translation
A transformer-based model pre-trained on multiple Indian languages using Masked Language Modeling and Back-Translation Language Modeling techniques.
AI4BHARAT
  • See Upvoters0
  • Downloads18
  • Views296
  • File Size0

AI4Bharat- 500 M - RomanSetu Multilingual Native-to-Roman Model
RomanSetu is a multilingual continual pretrained transformer model designed for transliteration across six Indic languages
AI4BHARAT
  • See Upvoters1
  • Downloads57
  • Views1,210
  • File Size0

AI4Bharat- 400 M - RomanSetu Multilingual Native-to-Roman Model
RomanSetu is a multilingual continual pretrained transformer model designed for transliteration across six Indic languages
AI4BHARAT
  • See Upvoters1
  • Downloads98
  • Views1,309
  • File Size0

AI4Bharat- 500 M - RomanSetu Multilingual Roman-to-Native Model
RomanSetu is a multilingual continual pretrained transformer model designed for transliteration across six Indic languages
AI4BHARAT
  • See Upvoters0
  • Downloads30
  • Views527
  • File Size0

AI4Bharat- 400 M - RomanSetu Multilingual Roman-to-Native Model
RomanSetu is a multilingual continual pretrained transformer model designed for transliteration across six Indic languages
AI4BHARAT
  • See Upvoters0
  • Downloads46
  • Views904
  • File Size0