Czert – Czech BERT-like Model for Language Representation

Transformers have made a significant breakthrough in natural language processing. These models are trained on large datasets and can handle multiple tasks. We compare monolingual and multilingual transformer models for semantic relatedness and verse retrieval. We leveraged data from the original QurSim dataset (Arabic) and used authentic multi-author translations in 22 languages to create a multilingual QurSim dataset, which we released for the research community. We evaluated the performance of monolingual and multilingual LLMs for Arabic and our results show that monolingual LLMs give better results for verse classification and matching verse retrieval. We incrementally built monolingual models with Arabic, English, and Urdu and multilingual models with all 22 languages supported by the multilingual paraphrase-MiniLM-L12-v2 model. Our results show improvement in classification accuracy with the incorporation of multilingual QurSim.

In this paper, we present our approach for sentiment classification on Spanish-English code-mixed social media data in the SemEval-2020 Task 9. We investigate performance of various pre-trained Transformer models by using different fine-tuning strategies. We explore both monolingual and multilingual models with the standard fine-tuning method. Additionally, we propose a custom model that we fine-tune in two steps: once with a language modeling objective, and once with a task-specific objective. Although two-step fine-tuning improves sentiment classification performance over the base model, the large multilingual XLM-RoBERTa model achieves best weighted F1-score with 0.537 on development data and 0.739 on test data. With this score, our team jupitter placed tenth overall in the competition.

Structured pruning has been extensively studied on monolingual pre-trained language models and is yet to be fully evaluated on their multilingual counterparts. This work investigates three aspects of structured pruning on multilingual pre-trained language models: settings, algorithms, and efficiency. Experiments on nine downstream tasks show several counter-intuitive phenomena: for settings, individually pruning for each language does not induce a better result; for algorithms, the simplest method performs the best; for efficiency, a fast model does not imply that it is also small. To facilitate the comparison on all sparsity levels, we present Dynamic Sparsification, a simple approach that allows training the model once and adapting to different model sizes at inference. We hope this work fills the gap in the study of structured pruning on multilingual pre-trained models and sheds light on future research.

Large pretrained masked language models have become state-of-the-art solutions for many NLP problems. While studies have shown that monolingual models produce better results than multilingual models, the training datasets must be sufficiently large. We trained a trilingual LitLat BERT-like model for Lithuanian, Latvian, and English, and a monolingual Est-RoBERTa model for Estonian. We evaluate their performance on four downstream tasks: named entity recognition, dependency parsing, part-of-speech tagging, and word analogy. To analyze the importance of focusing on a single language and the importance of a large training set, we compare created models with existing monolingual and multilingual BERT models for Estonian, Latvian, and Lithuanian. The results show that the newly created LitLat BERT and Est-RoBERTa models improve the results of existing models on all tested tasks in most situations.KeywordsNatural language processingBERTTransformersEstonianLatvianLithuanian

The advent of pre-trained language models has directed a new era of Natural Language Processing (NLP), enabling us to create powerful language models. Among these models, Transformer-based models like BERT have grown in popularity due to their cutting-edge effectiveness. However, these models heavily rely on resource-intensive languages, forcing other languages into multilingual models(mBERT). The two fundamental challenges with mBERT become significantly more challenging in a resource-constrained language like Bangla. It was trained on a limited and organized dataset and contained weights for all other languages. Besides, current research on other languages suggests that a language-specific BERT model will exceed multilingual ones. This paper introduces Bangla-BERT^a, a monolingual BERT model for the Bangla language. Despite the limited data available for NLP tasks in Bangla, we perform pre-training on the largest Bangla language model dataset, BanglaLM, which we constructed using 40 GB of text data. Bangla-BERT achieves the highest results in all datasets and vastly improves the state-of-the-art performance in binary linguistic classification, multilabel extraction, and named entity recognition, outperforming multilingual BERT and other previous research. The pre-trained model is assessed against several non-contextual models such as Bangla fasttext and word2vec the downstream tasks. Finally, this model is evaluated by transfer learning based on hybrid deep learning models such as LSTM, CNN, and CRF in NER, and it is observed that Bangla-BERT outperforms state-of-the-art methods. The proposed Bangla-BERT model is assessed by using benchmark datasets, including Banfakenews, Sentiment Analysis on Bengali News Comments, and Cross-lingual Sentiment Analysis in Bengali. Finally, it is concluded that Bangla-BERT surpasses all prior state-of-the-art results by 3.52%, 2.2%, and 5.3%.

The Sundanese language has over 32 million speakers worldwide, but the language has reaped little to no benefits from the recent advances in natural language understanding. Like other low-resource languages, the only alternative is to fine-tune existing multilingual models. In this paper, we pre-trained three monolingual Transformer-based language models on Sundanese data. When evaluated on a downstream text classification task, we found that most of our monolingual models outperformed larger multilingual models despite the smaller overall pre-training data. In the subsequent analyses, our models benefited strongly from the Sundanese pre-training corpus size and do not exhibit socially biased behavior. We released our models for other researchers and practitioners to use.

Syntactic parsing is one of the areas in Natural Language Processing. The development of large-scale multilingual language models has enabled cross-lingual parsing approaches, which allows us to develop parsers for languages that do not have treebanks available. However, these approaches rely on the assumption that languages share orthographic representations and lexical entries. In this article, we investigate methods for developing a dependency parser for Xibe, a low-resource language that is written in a unique script. We first investigate lexicalized monolingual dependency parsing experiments to examine the effectiveness of word, part-of-speech, and character embeddings as well as pre-trained language models. Results show that character embeddings can significantly improve performance, while pre-trained language models decrease performance since they do not recognize the Xibe script. We also train delexicalized monolingual models, which yield competitive results to the best lexicalized model. Since the monolingual models are trained on a very small training set, we also investigate lexicalized and delexicalized cross-lingual models. We use six closely related languages as source language, which cover a wide range of scripts. In this setting, the delexicalized models achieve higher performance than lexicalized models. A final experiment shows that we can increase performance of the cross-lingual model by combining source languages and selecting the most similar sentences to Xibe as training set. However, all cross-lingual parsing results are still considerably lower than the monolingual model. We attribute the low performance of cross-lingual methods to syntactic and annotation differences as well as to the impoverished input of Universal Dependency Part-of-Speech tags that the delexicalized model has access to.

In this paper, we describe our three submissions to the SIGMORPHON 2020 shared task 1 on grapheme-to-phoneme conversion for 15 languages. We experimented with a single multilingual Transformer model. We observed that the multilingual model achieves results on par with our separately trained monolingual models and is even able to avoid a few of the errors made by the monolingual models.

In this investigation, we propose a solution for the author’s gender identification task called AGI-P. This task has several real-world applications across different fields, such as marketing and advertising, forensic linguistics, sociology, recommendation systems, language processing, historical analysis, education, and language learning. We created a new dataset to evaluate our proposed method. The dataset is balanced in terms of gender using a random sampling method and consists of 1944 samples in total. We use accuracy as an evaluation measure and compare the performance of the proposed solution (AGI-P) against state-of-the-art machine learning classifiers and fine-tuned pre-trained multilingual language models such as DistilBERT, mBERT, XLM-RoBERTa, and Multilingual DEBERTa. In this regard, we also propose a customized fine-tuning strategy that improves the accuracy of the pre-trained language models for the author gender identification task. Our extensive experimental studies reveal that our solution (AGI-P) outperforms the well-known machine learning classifiers and fine-tuned pre-trained multilingual language models with an accuracy level of 92.03%. Moreover, the pre-trained multilingual language models, fine-tuned with the proposed customized strategy, outperform the fine-tuned pre-trained language models using an out-of-the-box fine-tuning strategy. The codebase and the corpus will be publicly available for scholars in this area upon publication.

In this work we explore to what extent multilingual models can be trained for one language and applied to a different one for the task of Multiple Choice Question Answering. We employ the RACE dataset to fine-tune both a monolingual and a multilingual models and apply these models to another different collections in different languages. The results show that both monolingual and multilingual models can be zero-shot transferred to a different dataset in the same language maintaining its performance. Besides, the multilingual model still performs good when it is applied to a different target language. Additionally, we find that exams that are more difficult to humans are harder for machines too. Finally, we advance the state-of-the-art for the QA4MRE Entrance Exams dataset in several languages.

This paper investigates how multilingual language models benefit low-resource languages through our submission to the WMT 2025 Low-Resource Indic Language Translation shared task.We explore whether languages from related families can effectively support translation for low-resource languages that were absent or underrepresented during model training.Using a quantized multilingual pretrained foundation model, we examine zero-shot translation capabilities and cross-lingual transfer effects across three language families: Tibeto-Burman, Indo-Aryan, and Austroasiatic.Our findings demonstrate that multilingual models failed to leverage linguistic similarities, particularly evidenced within the Tibeto-Burman family.The study provides insights into the practical feasibility of zero-shot translation for low-resource language settings and the role of language family relationships in multilingual model performance.

With the recent explosion in popularity of voice assistant devices, there is a growing interest in making them available to user populations in additional countries and languages. However, to provide the highest accuracy and best performance for specific user populations, most existing voice assistant models are developed individually for each region or language, which requires linear investment of effort. In this paper, we propose a general multilingual model framework for Natural Language Understanding (NLU) models, which can help bootstrap new language models faster and reduce the amount of effort required to develop each language separately. We explore how different deep learning architectures affect multilingual NLU model performance. Our experimental results show that these multilingual models can reach same or better performance compared to monolingual models across language-specific test data while require less effort in creating features and model maintenance.

The massive spread of false information on social media has become a global risk especially in a global pandemic situation like COVID-19. False information detection has thus become a surging research topic in recent months. In recent years, supervised machine learning models have been used to automatically identify false information in social media. However, most of these machine learning models focus only on the language they were trained on. Given the fact that social media platforms are being used in different languages, managing machine learning models for each and every language separately would be chaotic. In this research, we experiment with multilingual models to identify false information in social media by using two recently released multilingual false information detection datasets. We show that multilingual models perform on par with the monolingual models and sometimes even better than the monolingual models to detect false information in social media making them more useful in real-world scenarios.

We present MeiteiRoBERTa, the first publicly available monolingual RoBERTa-based language model for Meitei (Manipuri), a lowresource language spoken by over 1.8 million people in Northeast India.Trained from scratch on 76 million words of Meitei text in Bengali script, our model achieves a perplexity of 65.89, representing a 5.2 improvement over multilingual baselines mBERT (341.56) and MuRIL (355.65).We argue that Meitei's agglutinative morphology and complex word formation present typological challenges that multilingual models with broad language coverage fail to capture effectively.Through comprehensive evaluation on perplexity, tokenization efficiency, and semantic representation quality, we demonstrate that domain-specific pretraining significantly outperforms general-purpose multilingual models for low-resource languages.Our model exhibits superior semantic understanding with 0.769 similarity separation compared to 0.035 for mBERT and near-zero for MuRIL, despite MuRIL's better tokenization efficiency (fertility: 3.29 vs. 4.65).We publicly release the model, training code, and datasets to accelerate NLP research for Meitei and other underrepresented Northeast Indian languages.

Concept-to-text Natural Language Generation is the task of expressing an input meaning representation in natural language. Previous approaches in this task have been able to generalise to rare or unseen instances by relying on a delexicalisation of the input. However, this often requires that the input appears verbatim in the output text. This poses challenges in multilingual settings, where the task expands to generate the output text in multiple languages given the same input. In this paper, we explore the application of multilingual models in concept-to-text and propose Language Agnostic Delexicalisation, a novel delexicalisation method that uses multilingual pretrained embeddings, and employs a character-level post-editing model to inflect words in their correct form during relexicalisation. Our experiments across five datasets and five languages show that multilingual models outperform monolingual models in concept-to-text and that our framework outperforms previous approaches, especially in low resource conditions.