Open-ended Visual Question Answering on Covid19 (Accuracy)
49.43AccuracyT2R-R2
Evaluation Results
| Method | Links | |
|---|---|---|
| T2R-R2Scale=4B2026.06 | 49.43 | |
| T2R-R3Scale=4B2026.06 | 41.37 | |
| T2R-R1Scale=4B2026.06 | 40.56 | |
| T2R-R3Scale=2B2026.06 | 40.37 | |
| T2R-R1Scale=2B2026.06 | 39.43 | |
| Frozen TeacherScale=2B2026.06 | 31.88 | |
| T2R-R2Scale=2B2026.06 | 30.56 | |
| Judge RewardScale=4B2026.06 | 29.24 | |
| RLVR OnlyScale=4B2026.06 | 28.86 | |
| Frozen TeacherScale=4B2026.06 | 27.54 | |
| Base ModelScale=4B2026.06 | 26.22 | |
| Base ModelScale=2B2026.06 | 22.64 | |
| RLVR OnlyScale=2B2026.06 | 21.69 | |
| Judge RewardScale=2B2026.06 | 15.66 |