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

Safety and Helpfulness Evaluation on SPA-VL

87.17Safety Score

COPSD (Standard)

Updated 1mo ago

Evaluation Results

Method	Links
COPSD (Standard) 2026.06		87.17	89.15
SAFER-STEER 2026.03		85.8	79.17
TIS 2026.03		85.66	78.3
SFT 2026.03		85.66	78.3
COPSD (Hybrid) 2026.06		84.53	89.02
SAFER-STEER 2026.03		83.4	77.03
SPA-VL 2026.03		83.4	76.42
SFT 2026.03		81.7	73.02
GRPO 2026.06		76.98	73.38
MMDPO 2026.03		76.05	78.54
SPA-VL 2026.03		76.04	69.62
COPSD (Hybrid) 2026.06		74.72	72.52
SFT 2026.06		73.58	86.27
TIS 2026.03		72.45	61.32
SaFeR-VLM 2026.03		71.13	79.81
OPD 2026.06		70.19	85.6
COPSD (Standard) 2026.06		69.81	76.72
Qwen2.5VL-7B 2026.03		69.43	73.2
SaFeR-VLM 2026.03		68.87	73.21
MMDPO 2026.03		65.09	63.02
Safe-RLHF-V 2026.06		61.51	45.18
GRPO 2026.06		60.38	62.74
GRLHF-V 2026.03		59.81	68.24
OPD 2026.06		58.11	60.61
SFT 2026.06		51.7	51.91
Qwen3-VL-4B 2026.06		50	50
Qwen2.5-VL-7B 2026.06		50	50
Qwen2.5VL-3B 2026.03		48.68	53.2
Safe-RLHF-V 2026.06		47.17	20.15
SafeRLHF-V 2026.03		46.23	40.94
SafeRLHF-V 2026.03		43.21	33.58
GRLHF-V 2026.03		40.75	43.4