arXiv subjects

Aitor Soroa

Publications and source records attributed to Aitor Soroa.

At least 19 recordsLinked to original sources

arXiv2026

Towards Reliable Multilingual LLMs-as-a-Judge: An Empirical Study

Large language models (LLMs) are increasingly used for the automatic evaluation of generated text, yet most prior work focuses on English. Despite the growing demand for multilingual evaluation, extending LLM-based evaluators to multilingual settings remains challenging, particularly for low-resource languages and scenarios where in-domain data is scarce. This work explores several strategies for developing multilingual LLMs-as-a-judge, considering whether in-domain data is available for fine-tuning or not. We systematically analyze English, Spanish, and Basque, representing high-, mid-, and low-resource languages, considering instruction translation, monolingual versus multilingual supervision, and model size. For evaluation, we extend two existing meta-evaluation datasets to Basque and Spanish. Our results reveal key trade-offs: When in-domain data is available, fine-tuned smaller models can achieve performance comparable to proprietary models, whereas zero-shot evaluation with larger models proves more effective in out-of-domain settings. We also observe that fine-tuning on out-of-domain data can adversely affect model performance. These findings provide practical guidance for building efficient, reliable multilingual evaluation pipelines. The data and code are publicly available at hitz-zentroa/mJudge.

cs.CL↗

arXiv2025

Multilingual Medical Reasoning for Question Answering with Large Language Models

Large Language Models (LLMs) with reasoning capabilities have recently demonstrated strong potential in medical Question Answering (QA). Existing approaches are largely English-focused and primarily rely on distillation from general-purpose LLMs, raising concerns about the reliability of their medical knowledge. In this work, we present a method to generate multilingual reasoning traces based on medical knowledge extracted from Wikipedia. We produce 500k traces in English, Italian, and Spanish, using a retrieval-augmented generation approach over medical information from Wikipedia. The traces are generated to solve medical questions drawn from MedQA and MedMCQA, which we extend to Italian and Spanish. We test our pipeline in both in-domain and out-of-domain settings across Medical QA benchmarks, and demonstrate that our reasoning traces improve performance both when utilized via in-context learning (few-shot) and supervised fine-tuning, yielding state-of-the-art results among 8B-parameter LLMs. We believe that these resources can support the development of more transparent clinical decision-support tools in multilingual settings. We release the full suite of resources: reasoning traces, translated QA datasets, Medical-Wikipedia, and fine-tuned models.

cs.CL↗

arXiv2025

BERnaT: Basque Encoders for Representing Natural Textual Diversity

Language models depend on massive text corpora that are often filtered for quality, a process that can unintentionally exclude non-standard linguistic varieties, reduce model robustness and reinforce representational biases. In this paper, we argue that language models should aim to capture the full spectrum of language variation (dialectal, historical, informal, etc.) rather than relying solely on standardized text. Focusing on the Basque language, we construct new corpora combining standard, social media, and historical sources, and pre-train the BERnaT family of encoder-only models in three configurations: standard, diverse, and combined. We further propose an evaluation framework that separates Natural Language Understanding (NLU) tasks into standard and diverse subsets to assess linguistic generalization. Results show that models trained on both standard and diverse data consistently outperform those trained on standard corpora, improving performance across all task types without compromising standard benchmark accuracy. These findings highlight the importance of linguistic diversity in building inclusive, generalizable language models.

cs.CL↗

arXiv2025

Instructing Large Language Models for Low-Resource Languages: A Systematic Study for Basque

Instructing language models with user intent requires large instruction datasets, which are only available for a limited set of languages. In this paper, we explore alternatives to conventional instruction adaptation pipelines in low-resource scenarios. We assume a realistic scenario for low-resource languages, where only the following are available: corpora in the target language, existing open-weight multilingual base and instructed backbone LLMs, and synthetically generated instructions sampled from the instructed backbone. We present a comprehensive set of experiments for Basque that systematically study different combinations of these components evaluated on benchmarks and human preferences from 1,680 participants. Our conclusions show that target language corpora are essential, with synthetic instructions yielding robust models, and, most importantly, that using as backbone an instruction-tuned model outperforms using a base non-instructed model. Scaling up to Llama 3.1 Instruct 70B as backbone, our model comes near frontier models of much larger sizes for Basque, without using any Basque instructions. We release code, models, instruction datasets, and human preferences to support full reproducibility in future research on low-resource language adaptation. https://github.com/hitz-zentroa/latxa-instruct

cs.CL↗

arXiv2025

Conditioning LLMs to Generate Code-Switched Text

Code-switching (CS) is still a critical challenge in Natural Language Processing (NLP), due to the limited availability of large-scale, diverse CS datasets for robust training and evaluation. Despite recent advances, the capabilities and limitations of LLMs in handling CS are still not fully understood. In this work, we investigate the extent to which LLMs can be used in a framework for CS text generation, focusing on the English-Spanish language pair. Our proposed methodology consists of back-translating natural CS sentences into monolingual English, and using the resulting parallel corpus to fine-tune LLMs to turn monolingual sentences into CS. We thoroughly analyse the models' performance through a study on human preferences, a qualitative error analysis, an evaluation with popular reference-based metrics and LLM-based judgment. Results show that fine-tuning can be a key step to ensure that current LLMs consistently generate fluent code-switched text and that our methodology generates high-quality outputs, expanding research opportunities in CS communication. We find that traditional metrics do not correlate with human judgement when assessing the quality of the generated CS data, but LLM-based judgment aligns more closely with human preferences. We release our code and generated dataset under a CC-BY-NC-SA license.

cs.CL↗

arXiv2025

Euska\~nolDS: A Naturally Sourced Corpus for Basque-Spanish Code-Switching

Code-switching (CS) remains a significant challenge in Natural Language Processing (NLP), mainly due a lack of relevant data. In the context of the contact between the Basque and Spanish languages in the north of the Iberian Peninsula, CS frequently occurs in both formal and informal spontaneous interactions. However, resources to analyse this phenomenon and support the development and evaluation of models capable of understanding and generating code-switched language for this language pair are almost non-existent. We introduce a first approach to develop a naturally sourced corpus for Basque-Spanish code-switching. Our methodology consists of identifying CS texts from previously available corpora using language identification models, which are then manually validated to obtain a reliable subset of CS instances. We present the properties of our corpus and make it available under the name Euska\~nolDS.

cs.CL↗

arXiv2024

Latxa: An Open Language Model and Evaluation Suite for Basque

We introduce Latxa, a family of large language models for Basque ranging from 7 to 70 billion parameters. Latxa is based on Llama 2, which we continue pretraining on a new Basque corpus comprising 4.3M documents and 4.2B tokens. Addressing the scarcity of high-quality benchmarks for Basque, we further introduce 4 multiple choice evaluation datasets: EusProficiency, comprising 5,169 questions from official language proficiency exams; EusReading, comprising 352 reading comprehension questions; EusTrivia, comprising 1,715 trivia questions from 5 knowledge areas; and EusExams, comprising 16,774 questions from public examinations. In our extensive evaluation, Latxa outperforms all previous open models we compare to by a large margin. In addition, it is competitive with GPT-4 Turbo in language proficiency and understanding, despite lagging behind in reading comprehension and knowledge-intensive tasks. Both the Latxa family of models, as well as our new pretraining corpora and evaluation datasets, are publicly available under open licenses. Our suite enables reproducible research on methods to build LLMs for low-resource languages.

cs.CL↗

arXiv2024

A LLM-Based Ranking Method for the Evaluation of Automatic Counter-Narrative Generation

This paper proposes a novel approach to evaluate Counter Narrative (CN) generation using a Large Language Model (LLM) as an evaluator. We show that traditional automatic metrics correlate poorly with human judgements and fail to capture the nuanced relationship between generated CNs and human perception. To alleviate this, we introduce a model ranking pipeline based on pairwise comparisons of generated CNs from different models, organized in a tournament-style format. The proposed evaluation method achieves a high correlation with human preference, with a $\rho$ score of 0.88. As an additional contribution, we leverage LLMs as zero-shot CN generators and provide a comparative analysis of chat, instruct, and base models, exploring their respective strengths and limitations. Through meticulous evaluation, including fine-tuning experiments, we elucidate the differences in performance and responsiveness to domain-specific data. We conclude that chat-aligned models in zero-shot are the best option for carrying out the task, provided they do not refuse to generate an answer due to security concerns.

cs.CL↗

arXiv2024

BertaQA: How Much Do Language Models Know About Local Culture?

Large Language Models (LLMs) exhibit extensive knowledge about the world, but most evaluations have been limited to global or anglocentric subjects. This raises the question of how well these models perform on topics relevant to other cultures, whose presence on the web is not that prominent. To address this gap, we introduce BertaQA, a multiple-choice trivia dataset that is parallel in English and Basque. The dataset consists of a local subset with questions pertinent to the Basque culture, and a global subset with questions of broader interest. We find that state-of-the-art LLMs struggle with local cultural knowledge, even as they excel on global topics. However, we show that continued pre-training in Basque significantly improves the models' performance on Basque culture, even when queried in English. To our knowledge, this is the first solid evidence of knowledge transfer from a low-resource to a high-resource language. Our analysis sheds light on the complex interplay between language and knowledge, and reveals that some prior findings do not fully hold when reassessed on local topics. Our dataset and evaluation code are available under open licenses at https://github.com/juletx/BertaQA.

cs.CL↗

arXiv2024

XNLIeu: a dataset for cross-lingual NLI in Basque

XNLI is a popular Natural Language Inference (NLI) benchmark widely used to evaluate cross-lingual Natural Language Understanding (NLU) capabilities across languages. In this paper, we expand XNLI to include Basque, a low-resource language that can greatly benefit from transfer-learning approaches. The new dataset, dubbed XNLIeu, has been developed by first machine-translating the English XNLI corpus into Basque, followed by a manual post-edition step. We have conducted a series of experiments using mono- and multilingual LLMs to assess a) the effect of professional post-edition on the MT system; b) the best cross-lingual strategy for NLI in Basque; and c) whether the choice of the best cross-lingual strategy is influenced by the fact that the dataset is built by translation. The results show that post-edition is necessary and that the translate-train cross-lingual strategy obtains better results overall, although the gain is lower when tested in a dataset that has been built natively from scratch. Our code and datasets are publicly available under open licenses.

cs.CL↗

arXiv2024

Improving Explicit Spatial Relationships in Text-to-Image Generation through an Automatically Derived Dataset

Existing work has observed that current text-to-image systems do not accurately reflect explicit spatial relations between objects such as 'left of' or 'below'. We hypothesize that this is because explicit spatial relations rarely appear in the image captions used to train these models. We propose an automatic method that, given existing images, generates synthetic captions that contain 14 explicit spatial relations. We introduce the Spatial Relation for Generation (SR4G) dataset, which contains 9.9 millions image-caption pairs for training, and more than 60 thousand captions for evaluation. In order to test generalization we also provide an 'unseen' split, where the set of objects in the train and test captions are disjoint. SR4G is the first dataset that can be used to spatially fine-tune text-to-image systems. We show that fine-tuning two different Stable Diffusion models (denoted as SD$_{SR4G}$) yields up to 9 points improvements in the VISOR metric. The improvement holds in the 'unseen' split, showing that SD$_{SR4G}$ is able to generalize to unseen objects. SD$_{SR4G}$ improves the state-of-the-art with fewer parameters, and avoids complex architectures. Our analysis shows that improvement is consistent for all relations. The dataset and the code will be publicly available.

cs.CV↗

arXiv2023

Do Multilingual Language Models Think Better in English?

Translate-test is a popular technique to improve the performance of multilingual language models. This approach works by translating the input into English using an external machine translation system, and running inference over the translated input. However, these improvements can be attributed to the use of a separate translation system, which is typically trained on large amounts of parallel data not seen by the language model. In this work, we introduce a new approach called self-translate, which overcomes the need of an external translation system by leveraging the few-shot translation capabilities of multilingual language models. Experiments over 5 tasks show that self-translate consistently outperforms direct inference, demonstrating that language models are unable to leverage their full multilingual potential when prompted in non-English languages. Our code is available at https://github.com/juletx/self-translate.

cs.CL↗

arXiv2023

BLOOM: A 176B-Parameter Open-Access Multilingual Language Model

Large language models (LLMs) have been shown to be able to perform new tasks based on a few demonstrations or natural language instructions. While these capabilities have led to widespread adoption, most LLMs are developed by resource-rich organizations and are frequently kept from the public. As a step towards democratizing this powerful technology, we present BLOOM, a 176B-parameter open-access language model designed and built thanks to a collaboration of hundreds of researchers. BLOOM is a decoder-only Transformer language model that was trained on the ROOTS corpus, a dataset comprising hundreds of sources in 46 natural and 13 programming languages (59 in total). We find that BLOOM achieves competitive performance on a wide variety of benchmarks, with stronger results after undergoing multitask prompted finetuning. To facilitate future research and applications using LLMs, we publicly release our models and code under the Responsible AI License.

cs.CL↗

arXiv2023

Principled Paraphrase Generation with Parallel Corpora

Round-trip Machine Translation (MT) is a popular choice for paraphrase generation, which leverages readily available parallel corpora for supervision. In this paper, we formalize the implicit similarity function induced by this approach, and show that it is susceptible to non-paraphrase pairs sharing a single ambiguous translation. Based on these insights, we design an alternative similarity metric that mitigates this issue by requiring the entire translation distribution to match, and implement a relaxation of it through the Information Bottleneck method. Our approach incorporates an adversarial term into MT training in order to learn representations that encode as much information about the reference translation as possible, while keeping as little information about the input as possible. Paraphrases can be generated by decoding back to the source from this representation, without having to generate pivot translations. In addition to being more principled and efficient than round-trip MT, our approach offers an adjustable parameter to control the fidelity-diversity trade-off, and obtains better results in our experiments.

cs.CL↗

arXiv2023

The BigScience ROOTS Corpus: A 1.6TB Composite Multilingual Dataset

As language models grow ever larger, the need for large-scale high-quality text datasets has never been more pressing, especially in multilingual settings. The BigScience workshop, a 1-year international and multidisciplinary initiative, was formed with the goal of researching and training large language models as a values-driven undertaking, putting issues of ethics, harm, and governance in the foreground. This paper documents the data creation and curation efforts undertaken by BigScience to assemble the Responsible Open-science Open-collaboration Text Sources (ROOTS) corpus, a 1.6TB dataset spanning 59 languages that was used to train the 176-billion-parameter BigScience Large Open-science Open-access Multilingual (BLOOM) language model. We further release a large initial subset of the corpus and analyses thereof, and hope to empower large-scale monolingual and multilingual modeling projects with both the data and the processing tools, as well as stimulate research around this large multilingual corpus.

cs.CL↗

arXiv2022

Noisy Channel for Automatic Text Simplification

In this paper we present a simple re-ranking method for Automatic Sentence Simplification based on the noisy channel scheme. Instead of directly computing the best simplification given a complex text, the re-ranking method also considers the probability of the simple sentence to produce the complex counterpart, as well as the probability of the simple text itself, according to a language model. Our experiments show that combining these scores outperform the original system in three different English datasets, yielding the best known result in one of them. Adopting the noisy channel scheme opens new ways to infuse additional information into ATS systems, and thus to control important aspects of them, a known limitation of end-to-end neural seq2seq generative models.

cs.CL↗

arXiv2022

PoeLM: A Meter- and Rhyme-Controllable Language Model for Unsupervised Poetry Generation

Formal verse poetry imposes strict constraints on the meter and rhyme scheme of poems. Most prior work on generating this type of poetry uses existing poems for supervision, which are difficult to obtain for most languages and poetic forms. In this work, we propose an unsupervised approach to generate poems following any given meter and rhyme scheme, without requiring any poetic text for training. Our method works by splitting a regular, non-poetic corpus into phrases, prepending control codes that describe the length and end rhyme of each phrase, and training a transformer language model in the augmented corpus. During inference, we build control codes for the desired meter and rhyme scheme, and condition our language model on them to generate formal verse poetry. Experiments in Spanish and Basque show that our approach is able to generate valid poems, which are often comparable in quality to those written by humans.

cs.CL↗

arXiv2022

Does Corpus Quality Really Matter for Low-Resource Languages?

The vast majority of non-English corpora are derived from automatically filtered versions of CommonCrawl. While prior work has identified major issues on the quality of these datasets (Kreutzer et al., 2021), it is not clear how this impacts downstream performance. Taking representation learning in Basque as a case study, we explore tailored crawling (manually identifying and scraping websites with high-quality content) as an alternative to filtering CommonCrawl. Our new corpus, called EusCrawl, is similar in size to the Basque portion of popular multilingual corpora like CC100 and mC4, yet it has a much higher quality according to native annotators. For instance, 66% of documents are rated as high-quality for EusCrawl, in contrast with <33% for both mC4 and CC100. Nevertheless, we obtain similar results on downstream NLU tasks regardless of the corpus used for pre-training. Our work suggests that NLU performance in low-resource languages is not primarily constrained by the quality of the data, and other factors like corpus size and domain coverage can play a more important role.

cs.CL↗