arXiv subjects

Parameswari Krishnamurthy

Publications and source records attributed to Parameswari Krishnamurthy.

15 recordsLinked to original sources

arXiv2026

What Tokens are Learned when Tokenization is Optimized Jointly with Language Modeling?

Tokenization is a fundamental component of language modeling pipelines. Despite its importance, it is often fixed, even though it significantly impacts model performance across languages. In this work, we analyze what tokens are learned when tokenization is jointly optimized with language modeling. We compare tokenizer-free approaches such as SSLMs and H-Nets with fixed tokenizers across 18 typologically and script-diverse languages. Our results show that joint optimization fundamentally alters token structure. SSLMs recover morphologically aligned and contextually efficient tokens, whereas H-Nets prioritize byte-level efficiency, producing longer tokens with very low overlap with standard subword vocabularies. We further show that tokenization behavior varies across language typologies. Agglutinative languages exhibit more dynamic segmentation patterns while learning. Through downstream evaluation, with pretrained-then-finetuned BERT models, we find that SSLM-based pretokenization consistently reduces language modeling perplexity and achieves competitive downstream performance despite distinct vocabularies. Overall, tokenizer-free approaches optimize for contextual and computational efficiency rather than strict morphological structure, resulting in fundamentally different yet effective vocabularies for downstream NLP.

cs.CL↗

arXiv2026

SurakshaEval: An Indic Safety Benchmark for Multilingual LLMs

Existing safety evaluation datasets for large language models (LLMs) predominantly focus on English and Western contexts, often overlooking the linguistic diversity and culturally grounded safety risks present in other languages. To address this gap, we introduce SurakshaEval, a novel safety benchmark composed of human-written prompts spanning real-world scenarios, explicitly designed for ten major Indian languages - Assamese, Bengali, Gujarati, Hindi, Kannada, Malayalam, Marathi, Punjabi, Tamil, and Telugu, along with English. SurakshaEval includes both generic prompts common across India and region- and language-specific prompts that capture localized sociocultural sensitivities. We benchmark a broad range of state-of-the-art LLMs on SurakshaEval, establish baseline safety performance, and identify recurring failure modes, including over-refusal, missed detection of implicit bias, and insufficient contextual awareness in regionally sensitive settings. Our results show that even strong multilingual LLMs struggle to reliably meet nuanced safety requirements when operating in Indic languages, particularly in native scripts. These findings highlight the urgent need for safety evaluation frameworks that incorporate region-specific data and structured assessment protocols, enabling the development and deployment of AI systems that operate securely, ethically, and in alignment with diverse societal values. Our code and data are available at https://github.com/debobanerjee/SurakshaEval. Warning: This paper contains text that may be offensive or unsafe.

cs.CL↗

arXiv2026

Multilingual Idioms in Sentences and Conversations Across High-, Medium-, and Low-Resource Languages

Idiomatic expressions pose a major challenge for multilingual NLP because their meanings shift between figurative and literal usage, often requiring context for accurate interpretation. Prior work has focused on high-resource languages typically evaluates isolated idiom-meaning questions, overlooking realistic discourse. We introduce MIDI, a multilingual idiom dataset spanning 3 high-, 3 medium-, and 12 low-resource languages, curated by native speakers. Unlike previous datasets, MIDI provides idioms embedded in both sentence-level and conversational contexts, capturing both literal and figurative readings. Benchmarking state-of-the-art models shows that idiom comprehension degrades in low-resource languages and that, in all resource tiers, literal interpretations are substantially harder than figurative ones. Conversational context improves performance but does not eliminate these disparities. Through controlled tests and interventions on hidden representations, we further separate memorization from reasoning, exposing core limitations of current models.

cs.CL↗

arXiv2026

Typologically-Informed Candidate Reranking for LLM-based Translation into Low-Resource Languages

Large language models trained predominantly on high-resource languages exhibit systematic biases toward dominant typological patterns, leading to structural non-conformance when translating into typologically divergent low-resource languages. We present a framework that leverages linguistic typology to improve translation quality without parallel training data or model retraining. The framework consists of two components: the Universal Metalinguistic Framework (UMF), which represents languages as structured profiles across 16 typological dimensions with divergence-weighted scoring, and the Computational Engine, which operates through linguistic disambiguation during generation and typological compliance scoring during selection. Evaluation across nine language pairs demonstrates intervention rates strongly correlating with typological distance from English. In experiments on 341 English sentences each having different morphological and syntactic phenomena, the framework shows an intervention precision of 48.16% for conservatively treated languages, 28.15% for morphologically dense languages, and 86.26% for structurally profiled languages. The framework requires no parallel training data and operates with any LLM capable of producing multiple candidate outputs, enabling practical deployment for under-resourced languages.

cs.CL↗

arXiv2025

Decoding Fake Narratives in Spreading Hateful Stories: A Dual-Head RoBERTa Model with Multi-Task Learning

Social media platforms, while enabling global connectivity, have become hubs for the rapid spread of harmful content, including hate speech and fake narratives \cite{davidson2017automated, shu2017fake}. The Faux-Hate shared task focuses on detecting a specific phenomenon: the generation of hate speech driven by fake narratives, termed Faux-Hate. Participants are challenged to identify such instances in code-mixed Hindi-English social media text. This paper describes our system developed for the shared task, addressing two primary sub-tasks: (a) Binary Faux-Hate detection, involving fake and hate speech classification, and (b) Target and Severity prediction, categorizing the intended target and severity of hateful content. Our approach combines advanced natural language processing techniques with domain-specific pretraining to enhance performance across both tasks. The system achieved competitive results, demonstrating the efficacy of leveraging multi-task learning for this complex problem.

cs.CL↗

arXiv2025

Yes-MT's Submission to the Low-Resource Indic Language Translation Shared Task in WMT 2024

This paper presents the systems submitted by the Yes-MT team for the Low-Resource Indic Language Translation Shared Task at WMT 2024 (Pakray et al., 2024), focusing on translating between English and the Assamese, Mizo, Khasi, and Manipuri languages. The experiments explored various approaches, including fine-tuning pre-trained models like mT5 (Xue et al., 2020) and IndicBart (Dabre et al., 2021) in both multilingual and monolingual settings, LoRA (Hu et al., 2021) fine-tuning IndicTrans2 (Gala et al., 2023), zero-shot and few-shot prompting (Brown, 2020) with large language models (LLMs) like Llama 3 (Dubey et al., 2024) and Mixtral 8x7b (Jiang et al., 2024), LoRA supervised fine-tuning of Llama 3 (Mecklenburg et al., 2024), and training Transformer models (Vaswani, 2017) from scratch. The results were evaluated on the WMT23 Low-Resource Indic Language Translation Shared Task test data using SacreBLEU (Post, 2018) and CHRF (Popovic, 2015), highlighting the challenges of low-resource translation and the potential of LLMs for these tasks, particularly with fine-tuning.

cs.CL↗

arXiv2025

The Riddle of Reflection: Evaluating Reasoning and Self-Awareness in Multilingual LLMs using Indian Riddles

The extent to which large language models (LLMs) can perform culturally grounded reasoning across non-English languages remains underexplored. This paper examines the reasoning and self-assessment abilities of LLMs across seven major Indian languages-Bengali, Gujarati, Hindi, Kannada, Malayalam, Tamil, and Telugu. We introduce a multilingual riddle dataset combining traditional riddles with context-reconstructed variants and evaluate five LLMs-Gemini 2.5 Pro, Gemini 2.5 Flash, Mistral-Saba, LLaMA 4 Scout, and LLaMA 4 Maverick-under seven prompting strategies. In the first stage, we assess riddle-solving performance and find that while Gemini 2.5 Pro performs best overall, few-shot methods yield only marginal gains, and accuracy varies notably across languages. In the second stage, we conduct a self-evaluation experiment to measure reasoning consistency. The results reveal a key finding: a model's initial accuracy is inversely correlated with its ability to identify its own mistakes. Top-performing models such as Gemini 2.5 Pro are overconfident (4.34% True Negative Rate), whereas lower-performing models like LLaMA 4 Scout are substantially more self-aware (42.09% True Negative Rate). These results point to clear gaps in multilingual reasoning and highlight the need for models that not only reason effectively but also recognize their own limitations.

cs.CL↗

arXiv2025

Realizing LLMs' Causal Potential Requires Science-Grounded, Novel Benchmarks

Recent claims of strong performance by Large Language Models (LLMs) on causal discovery are undermined by a key flaw: many evaluations rely on benchmarks likely included in pretraining corpora. Thus, apparent success suggests that LLM-only methods, which ignore observational data, outperform classical statistical approaches. We challenge this narrative by asking: Do LLMs truly reason about causal structure, and how can we measure it without memorization concerns? Can they be trusted for real-world scientific discovery? We argue that realizing LLMs' potential for causal analysis requires two shifts: (P.1) developing robust evaluation protocols based on recent scientific studies to guard against dataset leakage, and (P.2) designing hybrid methods that combine LLM-derived knowledge with data-driven statistics. To address P.1, we encourage evaluating discovery methods on novel, real-world scientific studies. We outline a practical recipe for extracting causal graphs from recent publications released after an LLM's training cutoff, ensuring relevance and preventing memorization while capturing both established and novel relations. Compared to benchmarks like BNLearn, where LLMs achieve near-perfect accuracy, they perform far worse on our curated graphs, underscoring the need for statistical grounding. Supporting P.2, we show that using LLM predictions as priors for the classical PC algorithm significantly improves accuracy over both LLM-only and purely statistical methods. We call on the community to adopt science-grounded, leakage-resistant benchmarks and invest in hybrid causal discovery methods suited to real-world inquiry.

cs.LG↗

arXiv2025

CorIL: Towards Enriching Indian Language to Indian Language Parallel Corpora and Machine Translation Systems

India's linguistic landscape is one of the most diverse in the world, comprising over 120 major languages and approximately 1,600 additional languages, with 22 officially recognized as scheduled languages in the Indian Constitution. Despite recent progress in multilingual neural machine translation (NMT), high-quality parallel corpora for Indian languages remain scarce, especially across varied domains. In this paper, we introduce a large-scale, high-quality annotated parallel corpus covering 11 of these languages : English, Telugu, Hindi, Punjabi, Odia, Kashmiri, Sindhi, Dogri, Kannada, Urdu, and Gujarati comprising a total of 772,000 bi-text sentence pairs. The dataset is carefully curated and systematically categorized into three key domains: Government, Health, and General, to enable domain-aware machine translation research and facilitate effective domain adaptation. To demonstrate the utility of CorIL and establish strong benchmarks for future research, we fine-tune and evaluate several state-of-the-art NMT models, including IndicTrans2, NLLB, and BhashaVerse. Our analysis reveals important performance trends and highlights the corpus's value in probing model capabilities. For instance, the results show distinct performance patterns based on language script, with massively multilingual models showing an advantage on Perso-Arabic scripts (Urdu, Sindhi) while other models excel on Indic scripts. This paper provides a detailed domain-wise performance analysis, offering insights into domain sensitivity and cross-script transfer learning. By publicly releasing CorIL, we aim to significantly improve the availability of high-quality training data for Indian languages and provide a valuable resource for the machine translation research community.

cs.CL↗

arXiv2025

Rethinking Tokenization for Rich Morphology: The Dominance of Unigram over BPE and Morphological Alignment

The relationship between tokenizer algorithm (e.g., Byte-Pair Encoding (BPE), Unigram), morphological alignment, tokenization quality (e.g., compression efficiency), and downstream performance remains largely unclear, particularly for languages with complex morphology. In this paper, we conduct a comprehensive evaluation of tokenizers using small-sized BERT models -- from pre-training through fine-tuning -- for Telugu (agglutinative), along with preliminary evaluation in Hindi (primarily fusional with some agglutination) and English (fusional). To evaluate morphological alignment of tokenizers in Telugu, we create a dataset containing gold morpheme segmentations of 600 derivational and 7000 inflectional word forms. Our experiments reveal two key findings for Telugu. First, the choice of tokenizer algorithm is the most significant factor influencing performance, with Unigram-based tokenizers consistently outperforming BPE across most settings. Second, while better morphological alignment shows a moderate, positive correlation with performance on text classification and structure prediction tasks, its impact is secondary to the tokenizer algorithm. Notably, hybrid approaches that use morphological information for pre-segmentation significantly boost the performance of BPE, though not Unigram. Our results further showcase the need for comprehensive intrinsic evaluation metrics for tokenizers that could explain downstream performance trends consistently.

cs.CL↗

arXiv2025

Improving Genomic Models via Task-Specific Self-Pretraining

Pretraining DNA language models (DNALMs) on the full human genome is resource-intensive, yet often considered necessary for strong downstream performance. Inspired by recent findings in NLP and long-context modeling, we explore an alternative: self-pretraining on task-specific, unlabeled data. Using the BEND benchmark, we show that DNALMs trained with self-pretraining match or exceed the performance of models trained from scratch under identical compute. While genome-scale pretraining may still offer higher absolute performance, task-specific self-pretraining provides a practical and compute-efficient strategy for building stronger supervised baselines.

q-bio.GN↗

arXiv2025

keepitsimple at SemEval-2025 Task 3: LLM-Uncertainty based Approach for Multilingual Hallucination Span Detection

Identification of hallucination spans in black-box language model generated text is essential for applications in the real world. A recent attempt at this direction is SemEval-2025 Task 3, Mu-SHROOM-a Multilingual Shared Task on Hallucinations and Related Observable Over-generation Errors. In this work, we present our solution to this problem, which capitalizes on the variability of stochastically-sampled responses in order to identify hallucinated spans. Our hypothesis is that if a language model is certain of a fact, its sampled responses will be uniform, while hallucinated facts will yield different and conflicting results. We measure this divergence through entropy-based analysis, allowing for accurate identification of hallucinated segments. Our method is not dependent on additional training and hence is cost-effective and adaptable. In addition, we conduct extensive hyperparameter tuning and perform error analysis, giving us crucial insights into model behavior.

cs.CL↗

arXiv2024

Fine-tuning Pre-trained Named Entity Recognition Models For Indian Languages

Named Entity Recognition (NER) is a useful component in Natural Language Processing (NLP) applications. It is used in various tasks such as Machine Translation, Summarization, Information Retrieval, and Question-Answering systems. The research on NER is centered around English and some other major languages, whereas limited attention has been given to Indian languages. We analyze the challenges and propose techniques that can be tailored for Multilingual Named Entity Recognition for Indian Languages. We present a human annotated named entity corpora of 40K sentences for 4 Indian languages from two of the major Indian language families. Additionally,we present a multilingual model fine-tuned on our dataset, which achieves an F1 score of 0.80 on our dataset on average. We achieve comparable performance on completely unseen benchmark datasets for Indian languages which affirms the usability of our model.

cs.CL↗

arXiv2024

Exploring News Summarization and Enrichment in a Highly Resource-Scarce Indian Language: A Case Study of Mizo

Obtaining sufficient information in one's mother tongue is crucial for satisfying the information needs of the users. While high-resource languages have abundant online resources, the situation is less than ideal for very low-resource languages. Moreover, the insufficient reporting of vital national and international events continues to be a worry, especially in languages with scarce resources, like \textbf{Mizo}. In this paper, we conduct a study to investigate the effectiveness of a simple methodology designed to generate a holistic summary for Mizo news articles, which leverages English-language news to supplement and enhance the information related to the corresponding news events. Furthermore, we make available 500 Mizo news articles and corresponding enriched holistic summaries. Human evaluation confirms that our approach significantly enhances the information coverage of Mizo news articles. The mizo dataset and code can be accessed at \url{https://github.com/barvin04/mizo_enrichment

cs.CL↗

arXiv2023

Assessing Translation capabilities of Large Language Models involving English and Indian Languages

Generative Large Language Models (LLMs) have achieved remarkable advancements in various NLP tasks. In this work, our aim is to explore the multilingual capabilities of large language models by using machine translation as a task involving English and 22 Indian languages. We first investigate the translation capabilities of raw large language models, followed by exploring the in-context learning capabilities of the same raw models. We fine-tune these large language models using parameter efficient fine-tuning methods such as LoRA and additionally with full fine-tuning. Through our study, we have identified the best performing large language model for the translation task involving LLMs, which is based on LLaMA. Our results demonstrate significant progress, with average BLEU scores of 13.42, 15.93, 12.13, 12.30, and 12.07, as well as CHRF scores of 43.98, 46.99, 42.55, 42.42, and 45.39, respectively, using 2-stage fine-tuned LLaMA-13b for English to Indian languages on IN22 (conversational), IN22 (general), flores200-dev, flores200-devtest, and newstest2019 testsets. Similarly, for Indian languages to English, we achieved average BLEU scores of 14.03, 16.65, 16.17, 15.35 and 12.55 along with chrF scores of 36.71, 40.44, 40.26, 39.51, and 36.20, respectively, using fine-tuned LLaMA-13b on IN22 (conversational), IN22 (general), flores200-dev, flores200-devtest, and newstest2019 testsets. Overall, our findings highlight the potential and strength of large language models for machine translation capabilities, including for languages that are currently underrepresented in LLMs.

cs.CL↗