Multimodal Question Answering on SlideQA
17.1EMNemo
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Nemok=1, Generation VLM=LLaVA-1.5-13B2026.06 | 17.1 | 26.2 | 7.2 | 621 | |
| Latent Memoryk=1, Generation VLM=LLaVA-1.5-13B2026.06 | 11.1 | 15.7 | 21.5 | 37 | |
| Nemok=2, Generation VLM=LLaVA-1.5-13B2026.06 | 8.3 | 17.7 | 14.1 | 1,212 | |
| Latent Memoryk=2, Generation VLM=LLaVA-1.5-13B2026.06 | 7 | 12.4 | 33.6 | 47 | |
| Latent Memoryk=5, Generation VLM=LLaVA-1.5-13B2026.06 | 5.5 | 10.3 | 56 | 77 | |
| Nemok=5, Generation VLM=LLaVA-1.5-13B2026.06 | 4.7 | 13.2 | 30.5 | 2,887 | |
| Full ContextGeneration VLM=LLaVA-1.5-13B2026.06 | 0 | 0 | — | 11,871 |