List your company

Alignment and Safety Evaluation on Safety

73Avg@kBase
57.223261.319165.41569.5109May 8, 2026

Evaluation results

MethodLinks
BaseBackbone=Qwen2.5-7B-Instruct2026.057384.1
DFTBackbone=Qwen2.5-7B-Instruct2026.0572.3578.3
RPSFTBackbone=Qwen2.5-7B-Instruct2026.0570.4579.8
SFTBackbone=Qwen2.5-7B-Instruct2026.0570.383.3
IWBackbone=Qwen2.5-7B-Instruct2026.0569.880.8
DFTBackbone=Qwen2.5-3B-Instruct2026.0566.0368.2
BaseBackbone=Qwen2.5-3B-Instruct2026.0565.969.1
IWBackbone=Llama-3.1-8B-Instruct2026.0562.0873.1
RPSFTBackbone=Qwen2.5-3B-Instruct2026.0560.8376.2
RPSFTBackbone=Llama-3.1-8B-Instruct2026.0560.5568.6
DFTBackbone=Llama-3.1-8B-Instruct2026.0560.4564.6
IWBackbone=Qwen2.5-3B-Instruct2026.056069.7
SFTBackbone=Qwen2.5-3B-Instruct2026.0559.9880.2
BaseBackbone=Llama-3.1-8B-Instruct2026.0559.4376.6
SFTBackbone=Llama-3.1-8B-Instruct2026.0557.8367.8