Memes have become ubiquitous on social media platforms blending text and imagery to express complex and culturally nuanced messages. While a high degree of automation in meme moderation has been achieved for high-resource languages, low-resource languages, such as Nepali, still remain largely neglected. In this paper, we describe our system submission to the CHiPSAL 2026 Shared Task on Multi-modal Hate and Sentiment Understanding in Low-Resource Nepali Memes, which features two main sub-tasks: (1) Detection of HateSpeech as binary classification and (2) Sentiment Analysis as multi-class classification in Nepali memes. We perform a comprehensive analysis of the following models: uni-modal textual models (mBERT, XLM-RoBERTa,MuRIL), uni-modal visual models (ResNet, ConvNeXt, ViT), nine different late-fusion multimodal models, and the vision-language foundation model, SigLIP. Among all models, the ViT model achieved the best macro F1-score(0.6278) for the hate speech detection task, while SigLIP achieved the best score (0.5481) for the sentiment analysis task. We hypothesize that the under-performance of fusion models may be attributed to OCR noise and inadequate low-resource textual representations that act as a bottleneck when paired with more advanced visual encoders. These results highlight the unique challenges of multimodal meme comprehension in low-resource contexts and underscores the requirement for culturally grounded, noise-robust approaches to content moderation in Nepali.

The Devanagari script, an Indic script used by a diverse range of South Asian languages, presents a significant challenge in Natural Language Processing (NLP) research. The dialect and language variation, complex script features, and limited language-specific tools make development difficult. This shared task aims to address this challenge by bringing together researchers and practitioners to solve three key problems: Language identification, Hate speech detection, and Targets of Hate speech identification. The selected languages- Hindi, Nepali, Marathi, Sanskrit, and Bhojpuri- are widely used in South Asia and represent distinct linguistic structures. In this work, we explore the effectiveness of both machine-learning models and transformer-based models on all three sub-tasks. Our results demonstrate strong performance of the multilingual transformer model, particularly one pre-trained on domain-specific social media data, across all three tasks. The multilingual RoBERTa model, trained on the Twitter dataset, achieved a remarkable accuracy and F1-score of 99.5% on language identification (Task A), 88.3% and 72.5% on Hate Speech detection (Task B), and 68.6% and 61.8% on Hate Speech Target Classification (Task C).

In recent years, memes have developed as popular forms of online satire and critique, artfully merging entertainment, social critique, and political discourse. On the other side, memes have also become a medium for the spread of hate speech, misinformation, and bigotry, especially towards marginalized communities, including the LGBTQ+ population. Solving this problem calls for the development of advanced multimodal systems that analyze the complex interplay between text and visuals in memes. This paper describes our work in the CASE@RANLP 2025 shared task. As a part of that task, we developed systems for hate speech detection, target identification, stance classification, and humor recognition within the text of memes. We investigate two multimodal transformer-based systems, ResNet-18 with BERT and SigLIP2, for these sub-tasks. Our results show that SigLIP-2 consistently outperforms the baseline, achieving an F1 score of 79.27 in hate speech detection, 72.88 in humor classification, and competitive performance in stance 60.59 and target detection 54.86. Through this study, we aim to contribute to the development of ethically grounded, inclusive NLP systems capable of interpreting complex sociolinguistic narratives in multi-modal content.