@inproceedings{boudraa-etal-2025-implicit,
    title = "Implicit Hate Target Span Detection in Zero- and Few-Shot Settings with Selective Sub-Billion Parameter Models",
    author = "Boudraa, Hossam  and
      Favre, Benoit  and
      Urena, Raquel",
    editor = "Calabrese, Agostina  and
      de Kock, Christine  and
      Nozza, Debora  and
      Plaza-del-Arco, Flor Miriam  and
      Talat, Zeerak  and
      Vargas, Francielle",
    booktitle = "Proceedings of the The 9th Workshop on Online Abuse and Harms (WOAH)",
    month = aug,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/landing_page/2025.woah-1.21/",
    pages = "228--240",
    ISBN = "979-8-89176-105-6",
    abstract = "This work investigates the effectiveness of masked language models (MLMs) and autoregressive language models (LLMs) with fewer than one billion parameters in the detection of implicit hate speech through fine-grained span identification. The evaluation spans zero-shot, few-shot, and full supervision settings across two core benchmarks{---}SBIC and IHC{---}and an auxiliary testbed, OffensiveLang.RoBERTa-Large-355M emerges as the strongest zero-shot model, achieving the highest F1 scores of 75.8 (SBIC) and 72.5 (IHC), outperforming larger models like LLaMA 3.2-1B. ModernBERT-125M closely matches this performance with scores of 75.1 and 72.2, demonstrating the advantage of architectural efficiency. Among instruction-tuned models, SmolLM2-135M Instruct and LLaMA 3.2 1B Instruct consistently outperform their non-instructed counterparts, with up to +2.3 F1 gain on SBIC and +1.7 on IHC. Interestingly, the larger SmolLM2-360M Instruct does not outperform the 135M variant, highlighting that model scale does not always correlate with performance in implicit hate detection tasks.Few-shot fine-tuning with SmolLM2-135M Instruct achieves F1 scores of 68.2 (SBIC) and 64.0 (IHC), trailing full-data fine-tuning by only 1.6 and 2.0 points, respectively, with accuracy drops under 0.5 points. This illustrates the promise of compact, instruction-aligned models in data-scarce settings, particularly when optimized with Low-Rank Adaptation (LoRA).Topic-guided error analysis using Latent Dirichlet Allocation (LDA) reveals recurring model failures in ideologically charged or euphemistic discourse. Misclassifications often involve neutral references to identity, politics, or advocacy language, underscoring current limitations in discourse-level inference and sociopragmatic understanding."
}