List your company

Safety Evaluation on HEx-PHI Dolly risk-ranked

8.97S1 ASRLARF
5.728827.606949.48571.3631May 6, 2026

Evaluation results

MethodLinks
LARFBase Model=Llama2-7b-Chat2026.058.9716.931.3814.484.834.14—
Self-Inf-NBase Model=Llama3.1-8B-Instruct2026.0515.8657.9369.3166.2119.66-3.8—
Bi-Anchor(Grad)Base Model=Qwen3-8B2026.0519.6632.0733.4530.6921.38-1.72—
Self-Inf-NBase Model=Qwen3-8B2026.0519.6612.7622.4128.9726.21-6.55—
Self-Inf-NBase Model=Llama2-7b-Chat2026.0521.0313.4510.6920.6910.6910.34—
Bi-Anchor(Reps)Base Model=Llama2-7b-Chat2026.0525.8631.0318.974.8326.21-0.35—
Bi-Anchor(Reps)Base Model=Qwen3-8B2026.0531.3822.4126.2140.3427.593.79—
Bi-Anchor(Grad)Base Model=Llama2-7b-Chat2026.0533.4511.7213.17.2422.7610.69—
Bi-Anchor(Reps)Base Model=Llama3.1-8B-Instruct2026.0548.6248.6256.952.7616.931.72—
SQSD(Beaver)Base Model=Llama2-7b-Chat2026.0553.7930.6920.696.214.4849.31—
LARFBase Model=Qwen3-8B2026.0554.1444.4818.97105.5248.62—
Reward ModelBase Model=Llama2-7b-Chat2026.0554.1412.7612.761.721.3852.76—
SQSD(Aegis)Base Model=Llama2-7b-Chat2026.0558.9735.17208.284.8354.14—
Reward ModelBase Model=Qwen3-8B2026.0563.4530.3416.5515.179.3154.14—
SQSD(Aegis)Base Model=Qwen3-8B2026.0568.9747.2427.2414.1412.0756.9—
LARFBase Model=Llama3.1-8B-Instruct2026.057079.3131.3838.288.2861.72—
Bi-Anchor(Grad)Base Model=Llama3.1-8B-Instruct2026.0571.0366.5547.2422.7616.5554.48—
Reward ModelBase Model=Llama3.1-8B-Instruct2026.0581.3847.5925.1744.1424.4856.9—
SQSD(Beaver)Base Model=Qwen3-8B2026.0582.7641.3827.5917.9311.0371.73—
SQSD(Aegis)Base Model=Llama3.1-8B-Instruct2026.0582.7626.5552.4125.8610.6972.07—
SQSD(Beaver)Base Model=Llama3.1-8B-Instruct2026.059061.3842.078.622.7687.24—