List your company

Safety Classification on HoliSafe-Bench (ECE, NLL)

8.4ECELlama
6.80817.55428.339.046May 1, 2026

Evaluation results

MethodLinks
LlamaPrompt Selection Protocol=Mean2026.058.40.689
MiniCPMPrompt Selection Protocol=Mean2026.0514.40.658
Qwen3-VL 4BPrompt Selection Protocol=Mean2026.0515.30.685
MiniCPMPrompt Selection Protocol=Random2026.0515.30.73
InternVLPrompt Selection Protocol=Mean2026.0515.50.682
Qwen3-VL 8BPrompt Selection Protocol=Mean2026.0516.60.727
GemmaPrompt Selection Protocol=Mean2026.0521.50.902
InternVLPrompt Selection Protocol=Random2026.0521.71.082
Qwen30BPrompt Selection Protocol=Mean2026.0522.70.857
Qwen3-VL 4BPrompt Selection Protocol=Selected2026.0524.11.338
Qwen3-VL 8BPrompt Selection Protocol=Random2026.0524.21.805
InternVLPrompt Selection Protocol=Selected2026.0524.61.07
Qwen3-VL 4BPrompt Selection Protocol=Random2026.0524.81.716
Qwen30BPrompt Selection Protocol=Selected2026.0524.81.656
LlamaPrompt Selection Protocol=Random2026.0525.31.234
Qwen3-VL 8BPrompt Selection Protocol=Selected2026.0526.41.028
Qwen30BPrompt Selection Protocol=Random2026.0528.41.359
LlamaPrompt Selection Protocol=Selected2026.0528.41.074
GemmaPrompt Selection Protocol=Random2026.0534.82.769
MiniCPMPrompt Selection Protocol=Selected2026.0537.51.389
GemmaPrompt Selection Protocol=Selected2026.0548.22.292