Although machine translation is often considered solved for high-resource languages, it remains challenging for texts involving poetic language, philosophical concepts, and layered metaphors, characteristics that are central to Sanskrit literature. Additionally, Sanskrit’s rich morphology, sandhi, and compounding, combined with its multi-millennial and multi-domain textual tradition, highlight the need for large, diverse parallel resources, which are currently scarce. We introduce Mitrasaṃgraha, a large-scale Sanskrit-English machine translation dataset containing 391,548 aligned sentence pairs, over 4 times larger than the previously available Itihāsa corpus (Aralikatte et al., 2021). The dataset spans more than 3 millennia of Sanskrit literature across multiple domains and includes temporally annotated data for studying domain and period effects on translation performance. We release manually corrected development (5,587) and test (5,552) sets. Benchmarks show that fine-tuning models such as NLLB and Gemma substantially improves translation quality, and retrieval-augmented prompting further enhances performance. In a controlled cross-dataset comparison, training on Mitrasaṃgraha outperforms both Itihāsa and web-mined bitext for two model families. Despite these gains, our results reveal persistent challenges in translating complex compounds, philosophical terminology, and metaphorical language in Sanskrit, highlighting the need for further research in this area.

We present our approach to LT-EDI@ACL 2026 on counter-narrative generation for homophobic and transphobic comments. Generating high-quality counter-narratives in multilingual and low-resource settings remains challenging, particularly when data imbalance and script variation affect model performance. To address these issues, we explore multiple modeling strategies built around Gemma 3 12B with QLoRA fine-tuning, including data rebalancing and alternative input strategies for Tamil. Our findings show that task-specific fine-tuning combined with native-script Tamil produces more stable and higher-quality outputs than large few-shot prompts or transliteration-basedinputs. On the official leaderboard, our system ranks second in English with an overall score of 86.35% and sixth in Tamil with 63.77%,highlighting both the effectiveness of targeted fine-tuning and the challenges of low-resource counter-narrative generation.

This paper presents our system for the LT-EDI@ACL 2026 workshop on meme classification of homophobia and transphobia in English, Hindi, and Chinese. Detecting harmful content in memes is challenging because meaning often emerges from the interaction between visual elements and short textual cues, particularly in multilingual settings. To address this, we build a multimodal pipeline using CLIP ViT-L/14 visual embeddings, EasyOCR text extraction, TF–IDF lexical features, and a multinomial logistic regression classifier. We further incorporate two optional expert modules, a LoRA-adapted Qwen2-VL model and a CLIP zero-shot classifier, and combine predictions using weighted majority voting. The system is intentionally lightweight and reproducible, demonstrating that strong pretrained transfer features paired with explicit OCR can provide robust multilingual meme moderation without extensive fine-tuning. On the official leaderboard, our submission ranks 1st in Hindi, 3rd in English, and 5th in Chinese.

We describe the IHLC team’s submissionto the LT-EDI ACL 2026 Shared Task onGender-Inclusive Language Generation andCounterfactual/Counter Narrative Generation.Our English-only system applies an activationsteeringapproach combined with carefullyengineered prompt templates to producegender-neutral rewrites and empathetic counternarratives.We summarize system design, experimentalsetup, evaluation protocol used bythe shared task, and report results for both subtasks(Task A: Gender Inclusive Language Generation– Average = 80.00%, Rank 3; TaskB: Counter Narrative Generation – Average =78.12%, Rank 6). We also analyze strengthsand failure modes observed in automatic andhuman-checked evaluations and highlight directionsfor improvement.

We introduce grounded glossary generation, a structured task requiring models to recover semantically meaningful Sanskrit phrases and produce translation-grounded meanings from a śloka-translation pair, formalizing the traditional pāṭha-commentary practice as an evaluable NLP objective. We construct a benchmark of 31,316 śloka-translation-glossary triples from the Vālmīki Rāmāyaṇa and Śrīmad Bhāgavatam, paired with two metrics: Jaccard for phrase recovery and Meaning Faithfulness for semantic consistency. Across zero-shot, few-shot, and instruction fine-tuned variants of Gemma-3n-E4B, Gemma-3-12B, Phi-4, and Qwen3.5-9B, instruction fine-tuning substantially outperforms prompting, while explicit segmentation yields gains. Error analysis identifies over-segmentation of sandhi and samāsa compounds as the dominant failure mode, pointing to morphological modeling as the key bottleneck for faithful Sanskrit lexical decomposition.

Poetry generation in Sanskrit typically requires the verse to be semantically coherent and adhere to strict prosodic rules. In Sanskrit prosody, every line of a verse is typically a fixed length sequence of syllables adhering to prescribed binary patterns of syllable weights. We observe that instead of treating a verse as a monolithic sequence, segmenting them as grouped-lines leads to significant improvement in semantic coherence by 10% with comparable metrical adherence. Specifically, PINGALA, our proposed decoding approach is designed to encourage every line to have well-formed words and our token selection biases the model towards it by preferring longer tokens. Writing in Sanskrit follows phonemic orthography, hence using a phonetically aware transliteration scheme, SLP1, increased the metrical alignment by 46% with comparable semantic similarity, for a instruction fine-tuned large language models like Phi-4. We also introduce a new approach for reference-free evaluation using cross-encoders which achieved better alignment with true poetry instances.

Sanskrit multi-component compounding presents a severe structured prediction challenge: the space of valid parse structures grows exponentially with compound length, implicit semantic relations are highly context-sensitive, and accurate type assignment demands simultaneous global structural coherence and local morphological precision across a fine-grained label typology. Existing systems address Nested Compound Type Identification for Sanskrit (NeCTIS) discriminatively, predicting dependency edges and semantic types in a single greedy pass. Reformulating NeCTIS as iterative generative refinement via diffusion over the full label sequence resolves this structural inconsistency but exposes a complementary bottleneck - coarse global attention that suppresses the localized morphological signals necessary for fine-grained type discrimination. We address this through a local refinement mechanism within a restricted window.

We further provide the first systematic evaluation of large language models on NeCTIS, finding consistent and substantial failures under zero-shot and few-shot conditions - attributable to the absence of structural inductive biases, Sanskrit under-representation in pretraining, and morphological sensitivity requirements that broad-coverage models are fundamentally ill-equipped to satisfy.

Text Generation has achieved remarkable performance using large language models. It has also been recently well-studied that these large language models are capable of creative generation tasks but prominently for high-resource languages. This prompts a fundamental question: Is there a way to utilize these (large) language models for structured poetry generation in a low-resource language, such as Sanskrit? We present Chandomitra, an English input to structured Sanskrit Poetry translation dataset, specifically adhering to the Anushtubh meter. We benchmark various open and closed models, and scrutinize specialized techniques such as constrained decoding and instruction fine-tuning, for the proposed task. Our constrained decoding methodology achieves 99.86% syntactic accuracy in generating metrically valid Sanskrit poetry, outperforming GPT-4o (1-shot: 31.24%). Our best-performing instruction-tuned model, on the other hand, performs better in semantic coherence with the English input, at the expense of slightly lower syntactic accuracy. Human evaluation further reveals that instruction fine-tuned model is better able to capture the poetic aspects.

Large Language Models (LLMs) are increasingly treated as universal, general-purpose solutions across NLP tasks, particularly in English. But does this assumption hold for low-resource, morphologically rich languages such as Sanskrit? We address this question by comparing instruction-tuned and in-context-prompted LLMs with smaller task-specific encoder–decoder models on the Sanskrit poetry-to-prose conversion task. This task is intrinsically challenging: Sanskrit verse exhibits free word order combined with rigid metrical constraints, and its conversion to canonical prose (anvaya) requires multi-step reasoning involving compound segmentation, dependency resolution, and syntactic linearisation. This makes it an ideal testbed to evaluate whether LLMs can surpass specialised models.For LLMs, we apply instruction fine-tuning on general-purpose models and design in-context learning templates grounded in Pāṇinian grammar and classical commentary heuristics. For task-specific modelling, we fully fine-tune a ByT5-Sanskrit Seq2Seq model. Our experiments show that domain-specific fine-tuning of ByT5-Sanskrit significantly outperforms all instruction-driven LLM approaches. Human evaluation strongly corroborates this result, with scores exhibiting high correlation with Kendall’s Tau scores.Additionally, our prompting strategies provide an alternative to fine-tuning when domain-specific verse corpora are unavailable, and the task-specific Seq2Seq model demonstrates robust generalisation on out-of-domain evaluations.Our code¹ and dataset² are publicly available.

High lexical variation, ambiguous references, and long-range dependencies make entity resolution in literary texts particularly challenging. We present Mahānāma, the first large-scale dataset for end-to-end Entity Discovery and Linking (EDL) in Sanskrit, a morphologically rich and under-resourced language. Derived from the Mahābhārata , the world’s longest epic, the dataset comprises over 109K named entity mentions mapped to 5.5K unique entities, and is aligned with an English knowledge base to support cross-lingual linking. The complex narrative structure of Mahānāma, coupled with extensive name variation and ambiguity, poses significant challenges to resolution systems. Our evaluation reveals that current coreference and entity linking models struggle when evaluated on the global context of the test set. These results highlight the limitations of current approaches in resolving entities within such complex discourse. Mahānāma thus provides a unique benchmark for advancing entity resolution, especially in literary domains.

This overview paper presents the findings of the two shared tasks organized as part of the 1st Workshop on Benchmarks, Harmonization, Annotation, and Standardization for Human-Centric AI in Indian Languages (BHASHA) co-located with IJCNLP-AACL 2025. The shared tasks are: (1) Indic Grammar Error Correction (IndicGEC) and (2) Indic Word Grouping (IndicWG). For GEC, participants were tasked with producing grammatically correct sentences based on given input sentences in five Indian languages. For WG, participants were required to generate a word-grouped variant of a provided sentence in Hindi. The evaluation metric used for GEC was GLEU, while Exact Matching was employed for WG. A total of 14 teams participated in the final phase of the Shared Task 1; 2 teams participated in the final phase of Shared Task 2. The maximum GLEU scores obtained for Hindi, Bangla, Telugu, Tamil and Malayalam languages are respectively 85.69, 95.79, 88.17, 91.57 and 96.02 for the IndicGEC shared task. The highest exact matching score obtained for IndicWG shared task is 45.13%.