List your company

Content Moderation on Multi-category Chinese Content Moderation Dataset 1.0 (test)

89.7Politics AccuracyCARO (Llama3-8B)
8.89229.87150.8571.829Apr 12, 2026

Evaluation results

MethodLinks
CARO (Llama3-8B)Category=Our Model, Base Model=Llama3-8B2026.0489.794.597.481.497.272.388.8
CARO (Qwen2.5-7B)Category=Our Model, Base Model=Qwen2.5-7B2026.0489.593.597.881.298.474.689.2
Naive SFTCategory=Post-Training2026.0482.894.493.270.798.663.784.2
DeepSeek V3Category=General LLMs2026.047990.389.870.59562.580.3
CoT SFT+RLCategory=Post-Training2026.0478.3909172.395.466.381.6
CoT SFTCategory=Post-Training2026.0477.390.689.265.996.765.480
QwQ-32BCategory=General LLMs2026.0475.469.67260.784.954.669.1
Class-RAGCategory=Post-Training2026.0474.983.79168905675.5
DeepSeek R1Category=General LLMs2026.0472.791.486.164.694.359.777.1
Agentic ICLCategory=Post-Training2026.0466.88777.767.691.854.672.9
Qwen2.5-32B-InstructCategory=General LLMs2026.0459.191.184.467.995.454.274.3
GPT-4Category=General LLMs2026.0458.688.779.864.392.756.872.3
LLaMA3-8BCategory=General LLMs2026.0458.555.98165.290.644.267.5
Qwen2.5-7B-InstructCategory=General LLMs2026.0454.981.97060.184.348.864.3
LLaMA-Guard-3-8BCategory=Specific LLMs2026.041274.141.845.729.435.639.7