SZED at ArGuard Shared Tasks: Multi-Task Label-Attention and Cross-Modal Fusion for Arabic Hateful Meme Detection
arXiv:2610.26972v1 Announce Type: new Abstract: We present SZED's submission to ArGuard 2026 Track A: Arabic hateful meme detection. The task has two parts: binary classification (A1) and fine-grained multi-label tagging (A2). For A1, we fuse MARBERTv2 and CLIP ViT-L/14 embeddings, comparing concatenation, self-attention, and ensemble heads. For A2, we use a multi-task label-attention architecture with per-class query vectors and an auxiliary head for cascade calibration. We reach macro-F1 = 0.7
Read original source
Source record
Research analysis
Research claim
Paper identifier: 2610.26972. Source account: arXiv:2610.26972v1 Announce Type: new Abstract: We present SZED's submission to ArGuard 2026 Track A: Arabic hateful meme detection. The task has two parts: binary classification (A1) and fine-grained multi-label tagging (A2). For A1, we fuse MARBERTv2 and CLIP ViT-L/14 embeddings, comparing concatenation, self-attention, and ensemble heads. For A2, we use a multi-task label-attention architecture with per-class query vectors and an auxiliary head for cascade calibration. We reach macro-F1 = 0.7
Evaluation record
No structured result table, dataset, baseline or evaluation budget has been extracted. Any performance numbers in the source account remain attributed claims.
What the result supports
Interpret results only for the reported model version, task, data and inference budget. Community votes, downloads or a benchmark position do not establish reliability in a different workflow.
Validity gaps
Check sample selection, baseline configuration, uncertainty, contamination risk, negative results and access to code or weights. No reproduction by WeWorkBuddy is recorded here.
Verification plan
Read the paper or model card and its evaluation appendix. Reproduce one bounded result against the stated baseline, then test a representative user task and record where the conclusion fails.