List your company

Multi-turn MLLM Safety Evaluation on STEER-Beaver Multi-turn

55.58SafetySPA-VL
12.794423.902235.0146.1178Mar 18, 2026

Evaluation results

MethodLinks
SPA-VLBackbone=Qwen2.5VL-7B, Alignment Strategy=SPA-VL2026.0355.5877.34
SAFER-STEERBackbone=Qwen2.5VL-7B, Alignment Strategy=SAFER-STEER2026.0353.4475.57
SaFeR-VLMBackbone=Qwen2.5VL-7B, Alignment Strategy=SaFeR-VLM2026.0351.6958.26
SAFER-STEERBackbone=Qwen2.5VL-3B, Alignment Strategy=SAFER-STEER2026.0350.2771.2
SFTBackbone=Qwen2.5VL-7B, Alignment Strategy=SFT2026.035075.45
MMDPOBackbone=Qwen2.5VL-7B, Alignment Strategy=MMDPO2026.0340.7371.81
SFTBackbone=Qwen2.5VL-3B, Alignment Strategy=SFT2026.0339.4364.27
Qwen2.5VL-7BBackbone=Qwen2.5VL-7B, Alignment Strategy=Base2026.0336.0665.54
SPA-VLBackbone=Qwen2.5VL-3B, Alignment Strategy=SPA-VL2026.0334.953.35
GRLHF-VBackbone=Qwen2.5VL-7B, Alignment Strategy=GRLHF-V2026.0330.2355.01
TISBackbone=Qwen2.5VL-7B, Alignment Strategy=TIS2026.0329.2324.96
SafeRLHF-VBackbone=Qwen2.5VL-7B, Alignment Strategy=SafeRLHF-V2026.0328.4242.63
SaFeR-VLMBackbone=Qwen2.5VL-3B, Alignment Strategy=SaFeR-VLM2026.0326.2337.32
TISBackbone=Qwen2.5VL-3B, Alignment Strategy=TIS2026.0323.8426.16
MMDPOBackbone=Qwen2.5VL-3B, Alignment Strategy=MMDPO2026.0322.7142.43
Qwen2.5VL-3BBackbone=Qwen2.5VL-3B, Alignment Strategy=Base2026.0321.7942.88
GRLHF-VBackbone=Qwen2.5VL-3B, Alignment Strategy=GRLHF-V2026.0320.6339.86
SafeRLHF-VBackbone=Qwen2.5VL-3B, Alignment Strategy=SafeRLHF-V2026.0314.4423.17