lalalamdbf/AgentSafeEvo-Defender-Qwen3-8B
Text Generation • 8B • Updated • 8
AgentSafeEvo
AgentSafeEvo jointly evolves both attacker and defender to improve LLM agent safety.
-
-
lalalamdbf/AgentSafeEvo-Attacker-Qwen3-8B-RL
Text Generation • 8B • Updated • 4 -
lalalamdbf/AgentSafeEvo-Attacker-Qwen3-8B-SFT
Text Generation • 8B • Updated • 7 -
lalalamdbf/AgentSafeEvo-Attacker-SFT-Data
Viewer • Updated • 20.3k • 28 -
AgentSafeEvo: Adversarial Co-Evolution for LLM Agent Safety
Paper • 2610.19422 • Published • 5