# Low-resource Language Processing

A small-scale language model based on the BERT architecture, specifically designed for Kyrgyz natural language processing applications.

An abstract meaning representation parser fine-tuned on the Hungarian AMR dataset based on google/mt5-large

A BERT-based fine-tuned Turkish medical QA model specialized in extracting answers from medical texts

A universal Transformer model supporting over 100 languages, suitable for various natural language processing tasks

BnTQA-mBart is a low-resource Bengali table question answering model based on the mBART architecture, specifically designed for handling Bengali structured table data question answering tasks.

Turkish named entity recognition model fine-tuned based on dbmdz/bert-base-turkish-uncased

Urdu emotion recognition model fine-tuned on Whisper-medium, achieving 91.67% accuracy on the evaluation set

This model is fine-tuned based on the BERT architecture for detecting gibberish text (e.g., random character combinations) in Turkish.

This is an mBART model fine-tuned for English-to-Urdu translation tasks, based on the facebook/mbart-large-50 architecture and trained on a custom dataset.

A multilingual processing model supporting over 100 languages and writing systems, covering from mainstream languages to various dialects and minority languages

This is a Nepali sentence similarity calculation model based on sentence-transformers, which maps sentences and paragraphs into a 768-dimensional vector space.

This model is a multilingual summarization model based on the mT5 architecture, specifically fine-tuned for Sinhala to generate abstractive summaries of CNN Daily Mail Sinhala news.

Urdu text classification model based on XLM-RoBERTa architecture for binary sentiment analysis

A Malayalam news summarization model fine-tuned from google/mt5-small

This model is designed for the Bengali track of the SemEval Multiconer task, focusing on natural language processing tasks such as Named Entity Recognition (NER).

This is a model for the SemEval Multiconer task in the Bengali track, focusing on named entity recognition.

Marathi named entity recognition model fine-tuned from bert-base-multilingual-cased

IndicNER is a model specifically trained for recognizing named entities in sentences of 11 Indian languages, fine-tuned based on the bert-base-multilingual-uncased model.

Hindi Named Entity Recognition model based on MuRIL architecture, trained on the HiNER-original dataset

A BERT-base fine-tuned model for Indonesian text summarization, suitable for automatic summarization of Indonesian news articles

RobBERT is the state-of-the-art Dutch BERT model, pretrained on a large scale and adaptable to various text tasks through fine-tuning.

This is a named entity recognition model based on the XLM-RoBERTa model, fine-tuned on the Swahili portion of the MasakhaNER dataset.

This is a pre-trained named entity recognition model specifically designed for Hindi-English code-mixed data, trained on the LinCE dataset.

A multilingual pretrained model for Chinese and 7 minority languages in China

Gujarati automatic speech recognition model fine-tuned from facebook/wav2vec2-large-xlsr-53, achieving 23.55% WER on OpenSLR dataset

A Bengali automatic speech recognition model fine-tuned based on facebook/wav2vec2-large-xlsr-53, trained using OpenSLR dataset.

This is an automatic speech recognition (ASR) model fine-tuned on the Marathi speech dataset based on Facebook's wav2vec2-xls-r-300m model.

A Javanese masked language model based on DistilBERT, trained on Javanese Wikipedia

This is an automatic speech recognition model fine-tuned on Gujarati based on facebook/wav2vec2-large-xlsr-53, with a test WER of 28.92%.

A Persian Q&A generation model based on the GPT2 architecture, specifically designed for Persian Q&A tasks.

A Spanish abstract generation model fine-tuned based on google/mt5-base, excelling at extracting key information from text to generate concise summaries

A named entity recognition model fine-tuned on Bengali datasets based on sahajBERT, capable of identifying entity types such as person names, organization names, and location names.

This is a Transformer-based machine translation model for French (fr) to Igbo (ig), developed by the Helsinki-NLP team.

opus-mt-tw-fi is a machine translation model based on the transformer-align architecture, specifically designed for translating Ghana's Twi language (tw) into Finnish (fi).

This is a pretrained model specifically for the Khmer language, developed by Tsinghua University team, aiming to support natural language processing tasks in Khmer.

This is a pretrained model specifically for the Khmer language, developed by the Tsinghua University team.

This model is a Hausa speech recognition model fine-tuned on the Common Voice dataset based on facebook/wav2vec2-large-xlsr-53

opus-mt-fr-wls is a machine translation model based on the Transformer architecture, specifically designed for translating French (fr) to Wallisian (wls).

A Transformer-based neural machine translation model for French to American Sign Language (ASE), developed by the Helsinki-NLP team.

A Transformer-based machine translation model for English to Lunda, developed by the Helsinki-NLP team.