List your company

Code Security Evaluation on DiverseVul C/C++

142VULNS CountSelf-Play
36.9664.2391.5118.77Jun 2, 2026

Evaluation results

MethodLinks
Self-PlayBase Model=Qwen2.5-Coder-7B2026.06142
SFTBase Model=Qwen2.5-Coder-7B2026.06140
Self-PlayBase Model=Qwen2.5-Coder-3B2026.06130
Base LLMBase Model=CodeLlama-7B2026.06115
SFTBase Model=CodeLlama-7B2026.06110
SFTBase Model=Qwen2.5-Coder-3B2026.06107
Base LLMBase Model=Qwen2.5-Coder-7B2026.06106
Self-PlayBase Model=CodeLlama-7B2026.06103
TSP (OURS)Base Model=CodeLlama-7B2026.0694
Base LLMBase Model=Qwen2.5-Coder-3B2026.0693
TSP (OURS)Base Model=Qwen2.5-Coder-7B2026.0657
TSP (OURS)Base Model=Qwen2.5-Coder-3B2026.0641