List your company

Refusal on SaladBench

93.04ScoreICL
34.529649.719864.9180.1002Feb 7, 2026

Evaluation results

MethodLinks
ICLBackbone=Llama-3.1-8B-Instruct2026.0293.04
STEER2ADAPTBackbone=Llama-3.1-8B-Instruct2026.0291.84
REPBackbone=Llama-3.1-8B-Instruct2026.0290.46
STEER2ADAPTBackbone=Qwen-2.5-7B-Instruct2026.0288.52
ICLBackbone=Qwen-2.5-7B-Instruct2026.0287.32
Direct InferenceBackbone=Llama-3.1-8B-Instruct2026.0286.54
Few-ShotBackbone=Qwen-2.5-7B-Instruct, n=22026.0285.9
CAABackbone=Llama-3.1-8B-Instruct2026.0284.91
Few-ShotBackbone=Qwen-2.5-7B-Instruct, n=12026.0281.12
Direct InferenceBackbone=Qwen-2.5-7B-Instruct2026.0280.52
STEER2ADAPTBackbone=Mistral-7B-Instruct-v0.12026.0279.22
REPBackbone=Qwen-2.5-7B-Instruct2026.0279.21
ICLBackbone=Mistral-7B-Instruct-v0.12026.0275.89
REPBackbone=Mistral-7B-Instruct-v0.12026.0274.92
Few-ShotBackbone=Llama-3.1-8B-Instruct, n=12026.0274.24
CAABackbone=Qwen-2.5-7B-Instruct2026.0261.29
Few-ShotBackbone=Llama-3.1-8B-Instruct, n=22026.0255.29
CAABackbone=Mistral-7B-Instruct-v0.12026.0251.87
Direct InferenceBackbone=Mistral-7B-Instruct-v0.12026.0249.73
Few-ShotBackbone=Mistral-7B-Instruct-v0.1, n=22026.0236.99
Few-ShotBackbone=Mistral-7B-Instruct-v0.1, n=12026.0236.78