The NakbaVirality shared task focuses on multimodal virality prediction using a dataset of 2,600 multilingual posts collected from X and Reddit. In this work, we propose a multimodal architecture that combines XLM-RoBERTa for text encoding and a Vision Transformer (ViT) for image representation. The extracted features are aligned through bidirectional cross-attention to capture interactions between textual and visual modalities. To address the class imbalance present in the dataset, we apply focal loss, class weighting, and targeted data augmentation for the minority class. Additionally, layer-wise learning rate scheduling is used to stabilize fine-tuning of the pretrained encoders. Experimental results show that the proposed system achieves an accuracy of 0.6009 on the hidden test set, ranking 4th among 29 participating teams (107 total submissions). These results highlight the effectiveness of cross-modal attention mechanisms for modeling multimodal signals in high-stakes discourse.

We describe the Digilians submission to Task 1a (Spoken Visual Question Answering, MSA track) of the ImageEval 2026 shared task (Abdaljalil et al., 2026). Our system splits the problem into independently cacheable stages: speech recognition, schema-constrained transcript parsing that recovers a question and three candidate answers from continuous Arabic speech, a conditional repair mechanism that escalates to a larger ASR model only for semantically degenerate parses, and a joint multiplechoice visual reasoning stage that evaluates all three candidates against the image simultaneously rather than scoring them independently. On the labeled development split (500 records, synthetic voice-cloned audio), our production configuration reaches 79.0% accuracy against a 39.8% official baseline, with nearly the entire gain over an earlier configuration coming from a single change to the answer-selection model. On the blind final test split (1,000 records, human-recorded audio), the same frozen configuration scores 65.6% accuracy / 57.4% balanced accuracy / 50.4% macro F1, a substantial drop we attribute to the dev-to-test recordingcondition shift and report as our most important finding. We also surface a caching-related reproducibility issue uncovered during development and discuss what both findings imply for pipelines built the same way.