Hate Speech Detection on MHS
81.83AccuracyLLaMA-3.2-1B-LoRA
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| LLaMA-3.2-1B-LoRABackbone=LLaMA-3.2-1B-LoRA, Input=+Demo, Framework=Finetuned LLMs2026.05 | 81.83 | 78.28 | 76.17 | 77.08 | 87.54 | 0.49 | |
| gated demographic residual modelBackbone=BERTweet, Input=Ours, Framework=Finetuned PLMs2026.05 | 81.58 | 77.71 | 77.7 | 77.31 | 87.39 | 0.32 | |
| LLaMA-3.2-1B-LoRABackbone=LLaMA-3.2-1B-LoRA, Input=Text, Framework=Finetuned LLMs2026.05 | 81.17 | 77.34 | 75.67 | 76.4 | 87.05 | — | |
| gated demographic residual modelBackbone=ToxDect-RoBERTa, Input=Ours, Framework=Finetuned PLMs2026.05 | 80.93 | 76.81 | 77.21 | 76.99 | 87.02 | 0.16 | |
| BERTweetBackbone=BERTweet, Input=Text, Framework=Finetuned PLMs2026.05 | 80.8 | 76.8 | 76.94 | 76.74 | 87.07 | — | |
| ToxDect-RoBERTaBackbone=ToxDect-RoBERTa, Input=+Demo, Framework=Finetuned PLMs2026.05 | 80.79 | 76.79 | 76.01 | 76.31 | 86.75 | -0.11 | |
| Qwen2.5-0.5B-LoRABackbone=Qwen2.5-0.5B-LoRA, Input=+Demo, Framework=Finetuned LLMs2026.05 | 80.76 | 76.86 | 75.01 | 75.8 | 86.1 | 0.6 | |
| TID-DeBERTa-v3-baseBackbone=TID-DeBERTa-v3-base, Input=Text, Framework=Annotator modeling2026.05 | 80.59 | 76.54 | 76.45 | 76.39 | 86.82 | — | |
| BERTweetBackbone=BERTweet, Input=+Demo, Framework=Finetuned PLMs2026.05 | 80.58 | 76.53 | 77.95 | 77.05 | 87.22 | 0.15 | |
| TID-DeBERTa-v3-baseBackbone=TID-DeBERTa-v3-base, Input=+Demo, Framework=Annotator modeling2026.05 | 80.48 | 76.33 | 77.85 | 76.96 | 87 | 0.18 | |
| ToxDect-RoBERTaBackbone=ToxDect-RoBERTa, Input=Text, Framework=Finetuned PLMs2026.05 | 80.41 | 76.47 | 76.56 | 76.3 | 86.86 | — | |
| Qwen2.5-0.5B-LoRABackbone=Qwen2.5-0.5B-LoRA, Input=Text, Framework=Finetuned LLMs2026.05 | 80.23 | 76.12 | 74.48 | 75.2 | 85.5 | — | |
| LLaMA-3.2-1B-SPTBackbone=LLaMA-3.2-1B-SPT, Input=Text+Demo, Framework=Finetuned LLMs2026.05 | 79.55 | 75.24 | 74.15 | 74.6 | 84.5 | — | |
| LLaMA-2-7B-chat-hfBackbone=LLaMA-2-7B-chat-hf, Input=+Demo, Framework=Zero-shot2026.05 | 69.04 | 62.15 | 61.97 | 62.06 | 70.78 | -2.24 | |
| LLaMA-2-7B-chat-hfBackbone=LLaMA-2-7B-chat-hf, Input=Text, Framework=Zero-shot2026.05 | 40.97 | 64.98 | 58.12 | 39.39 | 73.02 | — | |
| Qwen2-7BBackbone=Qwen2-7B, Input=+Demo, Framework=Zero-shot2026.05 | 40.47 | 64.56 | 57.7 | 38.8 | 76.23 | -0.78 | |
| Qwen2-7BBackbone=Qwen2-7B, Input=Text, Framework=Zero-shot2026.05 | 40.06 | 64.48 | 57.42 | 38.29 | 77.01 | — | |
| Mistral-7B-v0.3Backbone=Mistral-7B-v0.3, Input=Text, Framework=Zero-shot2026.05 | 33.04 | 55.32 | 51.61 | 29.56 | 62.15 | — | |
| Mistral-7B-v0.3Backbone=Mistral-7B-v0.3, Input=+Demo, Framework=Zero-shot2026.05 | 29.04 | 55.09 | 50.05 | 22.67 | 63.93 | 1.78 |