Share your thoughts, 1 month free Claude Pro on usSee more

Safety Classification on HoliSafe-Bench (ECE, NLL)

8.4ECE

Llama

Updated 1mo ago

Evaluation Results

Method	Links
Llama 2026.05		8.4	0.689
MiniCPM 2026.05		14.4	0.658
Qwen3-VL 4B 2026.05		15.3	0.685
MiniCPM 2026.05		15.3	0.73
InternVL 2026.05		15.5	0.682
Qwen3-VL 8B 2026.05		16.6	0.727
Gemma 2026.05		21.5	0.902
InternVL 2026.05		21.7	1.082
Qwen30B 2026.05		22.7	0.857
Qwen3-VL 4B 2026.05		24.1	1.338
Qwen3-VL 8B 2026.05		24.2	1.805
InternVL 2026.05		24.6	1.07
Qwen3-VL 4B 2026.05		24.8	1.716
Qwen30B 2026.05		24.8	1.656
Llama 2026.05		25.3	1.234
Qwen3-VL 8B 2026.05		26.4	1.028
Qwen30B 2026.05		28.4	1.359
Llama 2026.05		28.4	1.074
Gemma 2026.05		34.8	2.769
MiniCPM 2026.05		37.5	1.389
Gemma 2026.05		48.2	2.292