Moving Direction Prediction on MoveDir
37AccuracyOriginal Model
Evaluation Results
| Method | Links | |
|---|---|---|
| Original ModelBackbone=LLaVA-1.5-13B, Retention Ratio=1.02025.12 | 37 | |
| TrimTokenator-LCBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.12 | 35.5 | |
| TrimTokenator-LCBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.12 | 34.5 | |
| DivPruneBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.12 | 34 | |
| DivPruneBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.12 | 34 | |
| VisionZipBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.12 | 33.5 | |
| VTWBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.12 | 33.5 | |
| SparseVLMBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.12 | 33.5 | |
| TrimTokenatorBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.12 | 33.5 | |
| FastVBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.12 | 33 | |
| VTWBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.12 | 32.5 | |
| CATPBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.12 | 32.5 | |
| FastVBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.12 | 32.5 | |
| VisionZipBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.12 | 32.5 | |
| SparseVLMBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.12 | 32 | |
| TrimTokenatorBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.12 | 32 | |
| Original ModelBackbone=LLaVA-1.5-7B, Retention Ratio=1.02025.12 | 32 | |
| CATPBackbone=LLaVA-1.5-13B, Retention Ratio=0.22025.12 | 32 | |
| TrimTokenator-LCBackbone=Yi-VL-6B, Retention Ratio=0.22025.12 | 30 | |
| DivPruneBackbone=Yi-VL-6B, Retention Ratio=0.22025.12 | 29.5 | |
| Original ModelBase Model=InternVL-Chat-ViT-6B-Vicuna-13B2025.12 | 29 | |
| SparseVLMBackbone=Yi-VL-6B, Retention Ratio=0.22025.12 | 29 | |
| TrimTokenatorBackbone=Yi-VL-6B, Retention Ratio=0.22025.12 | 29 | |
| FastVBackbone=Yi-VL-6B, Retention Ratio=0.22025.12 | 28.5 | |
| CATPBackbone=Yi-VL-6B, Retention Ratio=0.22025.12 | 28.5 | |
| VisionZipBackbone=Yi-VL-6B, Retention Ratio=0.22025.12 | 28 | |
| VTWBackbone=Yi-VL-6B, Retention Ratio=0.22025.12 | 27 | |
| TrimTokenator-LCBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.12 | 25.5 | |
| VisionZipBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.12 | 22.5 | |
| TrimTokenatorBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.12 | 22 | |
| Original ModelBackbone=Yi-VL-6B, Retention Ratio=1.02025.12 | 21 | |
| SparseVLMBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.12 | 20.5 | |
| VTWBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.12 | 19.5 | |
| FastVBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.12 | 18 | |
| DivPruneBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.12 | 17.5 | |
| CATPBackbone=LLaVA-1.5-7B, Retention Ratio=0.22025.12 | 17 |