VQA on Falcon-X (test)
99.93BLEUFalcon
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| FalconEvaluation Protocol=Fine-tuned2026.06 | 99.93 | 91.98 | 99.95 | 6.26 | |
| LISAEvaluation Protocol=Fine-tuned2026.06 | 99.91 | 91.42 | 99.93 | 6.25 | |
| GromaEvaluation Protocol=Fine-tuned2026.06 | 99.86 | 91.24 | 99.92 | 6.24 | |
| Sting-BeeEvaluation Protocol=Fine-tuned2026.06 | 99.12 | 91.27 | 99.92 | 6.24 | |
| LISAEvaluation Protocol=Zero-shot2026.06 | 3.86 | — | 3.94 | — | |
| Sting-BeeEvaluation Protocol=Zero-shot2026.06 | 3.86 | 7.74 | 4.04 | 0.01 | |
| GromaEvaluation Protocol=Zero-shot2026.06 | 3.32 | 8.9 | 3.44 | — | |
| Sa2VaEvaluation Protocol=Zero-shot2026.06 | 2.88 | — | 3.25 | — | |
| Kosmos-2Evaluation Protocol=Zero-shot2026.06 | 2.83 | 5.47 | 3.79 | 0.003 | |
| Llava-1.5Evaluation Protocol=Zero-shot2026.06 | 2.77 | 7.64 | 3.39 | — | |
| GLaMMEvaluation Protocol=Zero-shot2026.06 | 1.95 | 4.86 | 3.52 | 0.001 |