List your company

Generative Hallucination Evaluation on AMBER (test)

7.9CHAIR ScoreSID
3.6364.7435.856.957May 27, 2026

Evaluation results

MethodLinks
SIDModel=R1-Onevision2026.057.964.739.12.1
VCDModel=R1-Onevision2026.057.863.940.42
RLAIF-VModel=R1-Onevision2026.056.864.436.82.4
VanillaModel=R1-Onevision2026.056.663.735.62.1
OPA-DPOModel=R1-Onevision2026.056.662.932.81.8
OPA-DPOModel=MM-Eureka2026.055.157.521.51.6
VCDModel=MM-Eureka2026.054.861.524.31.4
SIDModel=MM-Eureka2026.054.660.723.71.1
VCDModel=ThinkLite-VL2026.054.658.120.21
OPA-DPOModel=ThinkLite-VL2026.054.661.525.21.8
RC-DPOModel=R1-Onevision2026.054.459.520.81.2
SIDModel=ThinkLite-VL2026.054.457.819.81
RLAIF-VModel=ThinkLite-VL2026.054.457.921.61.5
RLAIF-VModel=MM-Eureka2026.054.156181.1
VanillaModel=MM-Eureka2026.05459.321.21.2
VanillaModel=ThinkLite-VL2026.05458.318.31
RC-DPOModel=MM-Eureka2026.053.956.216.21
RC-DPOModel=ThinkLite-VL2026.053.856.815.21