List your company

Annotation on Annotation Task Suite Toxicity and Safety Domains (various)

84.4AccuracyGPT-4o-mini
71.81675.08378.3581.617May 30, 2026

Evaluation results

MethodLinks
GPT-4o-miniPrompting Condition=DSPy Opt.2026.0584.40.883
GPT-4o-miniPrompting Condition=DSPy Aligned2026.0584.30.883
GPT-4o-miniPrompting Condition=Few-Shot2026.0584.10.88
DeepSeek-V3Prompting Condition=FS + Def2026.0583.80.893
Qwen-2.5-72BPrompting Condition=Few-Shot2026.0583.80.894
DeepSeek-V3Prompting Condition=DSPy Aligned2026.0583.30.89
GPT-4o-miniPrompting Condition=Aligned Def2026.0583.30.875
GPT-4o-miniPrompting Condition=FS + Def2026.0583.30.88
Qwen-2.5-72BPrompting Condition=Zero-Shot2026.0583.30.876
Qwen-2.5-72BPrompting Condition=FS + Def2026.0583.20.892
Qwen-2.5-72BPrompting Condition=DSPy Aligned2026.0583.20.891
Llama-3.3-70BPrompting Condition=FS + Def2026.05830.904
DeepSeek-V3Prompting Condition=Aligned Def2026.05830.881
Mixtral-8x7BPrompting Condition=DSPy Opt.2026.0582.80.86
Mixtral-8x7BPrompting Condition=Aligned Def2026.0582.70.868
Mixtral-8x7BPrompting Condition=DSPy Aligned2026.0582.70.861
Llama-3.1-70BPrompting Condition=DSPy Aligned2026.0582.60.899
DeepSeek-V3Prompting Condition=Few-Shot2026.0582.60.89
Qwen-2.5-72BPrompting Condition=DSPy Opt.2026.0582.50.887
Mixtral-8x7BPrompting Condition=Few-Shot2026.0582.40.868
Llama-3.3-70BPrompting Condition=Aligned Def2026.0582.30.892
Mistral-Small-24BPrompting Condition=FS + Def2026.0582.30.879
GPT-4o-miniPrompting Condition=Overall2026.0582.30.862
Qwen-2.5-72BPrompting Condition=Aligned Def2026.0582.20.879
Qwen-2.5-72BPrompting Condition=Overall2026.0582.20.872
Llama-3.1-70BPrompting Condition=Aligned Def2026.0582.10.888
Llama-3.1-70BPrompting Condition=FS + Def2026.0582.10.898
Llama-3.3-70BPrompting Condition=Few-Shot2026.05820.896
Mistral-7BPrompting Condition=DSPy Opt.2026.05820.827
Mistral-7BPrompting Condition=DSPy Aligned2026.0581.70.825
Mixtral-8x7BPrompting Condition=Overall2026.0581.60.85
DeepSeek-V3Prompting Condition=Overall2026.0581.60.865
GPT-4o-miniPrompting Condition=Zero-Shot2026.0581.60.871
Mistral-7BPrompting Condition=Few-Shot2026.0581.50.852
Mistral-7BPrompting Condition=Misaligned2026.0581.40.823
Llama-3.3-70BPrompting Condition=DSPy Aligned2026.0581.30.904
Mistral-7BPrompting Condition=Aligned Def2026.0581.30.787
DeepSeek-V3Prompting Condition=Zero-Shot2026.0581.30.874
Llama-3.1-70BPrompting Condition=Few-Shot2026.0581.20.892
Mistral-7BPrompting Condition=Overall2026.0581.20.819
Mistral-Small-24BPrompting Condition=DSPy Aligned2026.0581.20.876
Mixtral-8x7BPrompting Condition=Misaligned2026.0581.20.843
Qwen-2.5-72BPrompting Condition=Misaligned2026.0581.20.868
GPT-4o-miniPrompting Condition=Misaligned2026.0581.10.857
Mistral-Small-24BPrompting Condition=Aligned Def2026.05810.863
Mixtral-8x7BPrompting Condition=Zero-Shot2026.05810.868
Mixtral-8x7BPrompting Condition=FS + Def2026.0580.90.859
DeepSeek-V3Prompting Condition=DSPy Opt.2026.0580.90.877
Mistral-7BPrompting Condition=FS + Def2026.0580.80.838
Mistral-Small-24BPrompting Condition=Few-Shot2026.0580.80.87
Mistral-Small-24BPrompting Condition=Misaligned2026.0580.70.85
DeepSeek-V3Prompting Condition=Misaligned2026.0580.70.856
Llama-3.3-70BPrompting Condition=Zero-Shot2026.0580.50.885
Mistral-7BPrompting Condition=Zero-Shot2026.0580.50.858
Mistral-Small-24BPrompting Condition=Overall2026.0580.50.852
Llama-3.1-8BPrompting Condition=Aligned Def2026.0580.10.843
Llama-3.3-70BPrompting Condition=Overall2026.0580.10.874
Llama-3.1-70BPrompting Condition=DSPy Opt.2026.0579.90.889
Llama-3.1-70BPrompting Condition=Zero-Shot2026.0579.80.881
Llama-3.1-70BPrompting Condition=Overall2026.0579.70.869
Mistral-Small-24BPrompting Condition=DSPy Opt.2026.0579.30.867
Llama-3.3-70BPrompting Condition=Misaligned2026.05790.863
Llama-3.1-70BPrompting Condition=Misaligned2026.0578.10.859
Mistral-Small-24BPrompting Condition=Zero-Shot2026.05780.868
Llama-3.3-70BPrompting Condition=DSPy Opt.2026.0577.50.884
Llama-3.1-8BPrompting Condition=Misaligned2026.05770.822
Llama-3.1-8BPrompting Condition=Overall2026.0576.80.818
Llama-3.1-8BPrompting Condition=Zero-Shot2026.0576.70.832
Llama-3.1-8BPrompting Condition=FS + Def2026.0575.40.824
Llama-3.1-8BPrompting Condition=Few-Shot2026.0575.30.822
Llama-3.1-8BPrompting Condition=DSPy Aligned2026.0573.50.821
Llama-3.1-8BPrompting Condition=DSPy Opt.2026.0572.30.809