List your company

Safety Evaluation on MalwareBench

3.02Attack Success Rate (ASR)CodeShield
-0.196821.516643.2364.9434Jun 10, 2026

Evaluation results

MethodLinks
CodeShieldBackbone=Qwen2.5-Coder-7B, Attack Configuration=w/o CodeSpear2026.063.021.04
CodeShieldBackbone=Qwen2.5-7B, Attack Configuration=w/ CodeSpear2026.063.330.62
CodeShieldBackbone=Qwen2.5-Coder-7B, Attack Configuration=w/ CodeSpear2026.063.441.15
CodeShieldBackbone=LLaMA3-8B, Attack Configuration=w/o CodeSpear2026.064.691.56
Safe-DPOBackbone=Qwen2.5-Coder-7B, Attack Configuration=w/o CodeSpear2026.067.155.21
Safe-DPOBackbone=LLaMA3-8B, Attack Configuration=w/o CodeSpear2026.067.54.06
CodeShieldBackbone=LLaMA3-8B, Attack Configuration=w/ CodeSpear2026.069.693.02
CodeShieldBackbone=Qwen2.5-7B, Attack Configuration=w/o CodeSpear2026.0621.566.98
VanillaBackbone=Qwen2.5-Coder-7B, Attack Configuration=w/o CodeSpear2026.0629.7920.83
Safe-DPOBackbone=Qwen2.5-7B, Attack Configuration=w/o CodeSpear2026.0632.414.58
Safe-DPOBackbone=Qwen2.5-7B, Attack Configuration=w/ CodeSpear2026.0638.1219.69
Safe-DPOBackbone=LLaMA3-8B, Attack Configuration=w/ CodeSpear2026.0644.3814.06
VanillaBackbone=LLaMA3-8B, Attack Configuration=w/o CodeSpear2026.0648.6533.96
VanillaBackbone=Qwen2.5-7B, Attack Configuration=w/o CodeSpear2026.0669.5839.38
VanillaBackbone=LLaMA3-8B, Attack Configuration=w/ CodeSpear2026.0670.126.98
Safe-DPOBackbone=Qwen2.5-Coder-7B, Attack Configuration=w/ CodeSpear2026.0676.5638.23
VanillaBackbone=Qwen2.5-7B, Attack Configuration=w/ CodeSpear2026.0683.0247.6
VanillaBackbone=Qwen2.5-Coder-7B, Attack Configuration=w/ CodeSpear2026.0683.4446.15