Open-ended Caption Generation on MSCOCO 2014 (val)
44CIDEr Scorevanilla
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| vanillaModel=MM-Eureka-7B2026.02 | 44 | 8.6 | |
| vanillaModel=ThinkLite-7B2026.02 | 41.7 | 7.8 | |
| OPA-DPOModel=Orsta-R1-7B2026.02 | 41.7 | 8.2 | |
| RLAIF-VModel=MM-Eureka-7B2026.02 | 41.3 | 8.5 | |
| OPA-DPOModel=MM-Eureka-7B2026.02 | 41.3 | 7.7 | |
| vanillaModel=Orsta-R1-7B2026.02 | 40.7 | 8.5 | |
| RLAIF-VModel=ThinkLite-7B2026.02 | 40.3 | 7.5 | |
| OPA-DPOModel=ThinkLite-7B2026.02 | 40 | 7.9 | |
| C3POModel=MM-Eureka-7B2026.02 | 40 | 7.7 | |
| RLAIF-VModel=Orsta-R1-7B2026.02 | 39.7 | 7.6 | |
| RLAIF-VModel=R1-Onevision-7B2026.02 | 39 | 8.3 | |
| RLAIF-VModel=MM-R1-7B2026.02 | 38.7 | 7.5 | |
| C3POModel=ThinkLite-7B2026.02 | 38 | 7.3 | |
| vanillaModel=MM-R1-7B2026.02 | 37 | 7.4 | |
| vanillaModel=R1-Onevision-7B2026.02 | 36.7 | 8.2 | |
| OPA-DPOModel=R1-Onevision-7B2026.02 | 36.7 | 7.7 | |
| OPA-DPOModel=MM-R1-7B2026.02 | 36.3 | 7.2 | |
| C3POModel=Orsta-R1-7B2026.02 | 35.3 | 6.9 | |
| C3POModel=R1-Onevision-7B2026.02 | 35 | 7.4 | |
| C3POModel=MM-R1-7B2026.02 | 35 | 7.2 |