Our new X account is live! Follow @wizwand_team for updates

Step-level tool invocation safety detection on ASB-Traj

0.9497Accuracy

TS-Guard

Updated 4d ago

Evaluation Results

Method	Links
TS-Guard 2026.01		0.9497	0.9476	0.9385
TS-Guard 2026.01		0.9064	0.7738	0.7824
TS-Guard 2026.01		0.8661	0.8328	0.8322
gpt-4o 2026.01		0.8295	0.5155	0.4435
Qwen3-8B 2026.01		0.8148	0.4066	0.3103
Qwen3Guard-8B-Gen 2026.01		0.7989	0.033	0.0168
Qwen2.5-7B-IT 2026.01		0.7928	0.2386	0.1589
Llama-Guard-3-8B 2026.01		0.767	0.2789	0.2203
Safiron 2026.01		0.7065	0.4397	0.563
ShieldAgent-THU 2026.01		0.6641	0.4834	0.7684
GPT-4o 2026.01		0.6584	0.6303	0.6011
Qwen3-8B 2026.01		0.5966	0.3893	0.2654
ShieldAgent-THU 2026.01		0.5763	0.5444	0.5227
Qwen3-8B 2026.01		0.5494	0.4028	0.4275
gpt-4o 2026.01		0.5492	0.5057	0.4939
Llama-Guard-3-8B 2026.01		0.5467	0.2482	0.1545
Qwen3Guard-8B-Gen 2026.01		0.535	0.1362	0.0757
Qwen2.5-7B-IT 2026.01		0.521	0.6296	0.8407
Safiron 2026.01		0.5163	0.3982	0.3303
Qwen2.5-7B-IT 2026.01		0.3585	0.3254	0.3791