Memorability Feedback Generation on MemBench
0.85IRINTERNVL3.5
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| INTERNVL3.5Category=Teacher oracles2026.02 | 0.85 | 11.92 | 2.4 | |
| QWEN2.5VLCategory=Teacher oracles2026.02 | 0.83 | 10.16 | 2.34 | |
| IDEFICS3Category=Teacher oracles2026.02 | 0.8 | 9.84 | 29.21 | |
| MemCoachCategory=Zero-shot baselines2026.02 | 0.8 | 7.21 | 4.99 | |
| GPT-5 MINICategory=Zero-shot baselines2026.02 | 0.75 | 7.03 | — | |
| LLAVA-OVCategory=Teacher oracles2026.02 | 0.74 | 5.93 | 5.73 | |
| AESEXPERTCategory=Aesthetics specialized2026.02 | 0.73 | 6.67 | 5.97 | |
| Q-INSTRUCTCategory=Aesthetics specialized2026.02 | 0.73 | 5.31 | 5.36 | |
| IDEFICS3Category=Zero-shot baselines2026.02 | 0.73 | 6.64 | 20.19 | |
| INTERNVL3.5Category=Zero-shot baselines2026.02 | 0.73 | 5.47 | 5.49 | |
| LLAVA-OVCategory=Zero-shot baselines2026.02 | 0.7 | 5.87 | 7.58 | |
| Edit modelCategory=Edit model2026.02 | 0.68 | 3.72 | — | |
| QWEN2.5VLCategory=Zero-shot baselines2026.02 | 0.68 | 4.26 | 10.23 |