Multimodal Understanding on SEED-Bench (val)
58.8AccuracyZOO-Prune
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ZOO-PruneBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 58.8 | — | — | — | — | — | — | — | — | — | — | |
| DivPruneBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 58.71 | — | — | — | — | — | — | — | — | — | — | |
| LLaVA-1.5-7BBackbone=LLaVA-1.5-7B, Retained Tokens=5762025.09 | 58.6 | — | — | — | — | — | — | — | — | — | — | |
| ZOO-PruneBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 57.53 | — | — | — | — | — | — | — | — | — | — | |
| FastVBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 57.1 | — | — | — | — | — | — | — | — | — | — | |
| DivPruneBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 56.98 | — | — | — | — | — | — | — | — | — | — | |
| VisionZipBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 56.4 | — | — | — | — | — | — | — | — | — | — | |
| FastVBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 55.9 | — | — | — | — | — | — | — | — | — | — | |
| ZOO-PruneBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 55.84 | — | — | — | — | — | — | — | — | — | — | |
| SparseVLMBackbone=LLaVA-1.5-7B, Retained Tokens=1922025.09 | 55.8 | — | — | — | — | — | — | — | — | — | — | |
| DivPruneBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 55.13 | — | — | — | — | — | — | — | — | — | — | |
| VisionZipBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 54.9 | — | — | — | — | — | — | — | — | — | — | |
| SparseVLMBackbone=LLaVA-1.5-7B, Retained Tokens=1282025.09 | 53.4 | — | — | — | — | — | — | — | — | — | — | |
| VisionZipBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 52.2 | — | — | — | — | — | — | — | — | — | — | |
| FastVBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 51.9 | — | — | — | — | — | — | — | — | — | — | |
| SparseVLMBackbone=LLaVA-1.5-7B, Retained Tokens=642025.09 | 51.1 | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-0.5b-instructBackbone=Qwen2.5-0.5b-instruct2026.05 | — | 71 | 42.9 | 40.2 | 71.8 | 50.1 | 44.8 | 66 | 67.4 | 61 | 60.5 | |
| Qwen2.5-0.5b-instruct+LngramBackbone=Qwen2.5-0.5b-instruct, Memory=Lngram2026.05 | — | 73.1 | 42.9 | 39.9 | 72 | 50.5 | 46.1 | 61.9 | 67.4 | 62.8 | 61.2 |