Text-rich Visual Question Answering on TextVQA (Accuracy)
80.07AccuracyVanilla
Evaluation Results
| Method | Links | |
|---|---|---|
| VanillaBackbone=Qwen3-VL-8B, TFLOPs=100%2026.06 | 80.07 | |
| Op-SkipBackbone=Qwen3-VL-8B, TFLOPs=66%, N=202026.06 | 78.45 | |
| APETBackbone=Qwen3-VL-8B, TFLOPs=66%2026.06 | 78.36 | |
| VanillaBackbone=Qwen2.5-VL-7B, TFLOPs=100%2026.06 | 77.77 | |
| V2DropBackbone=Qwen3-VL-8B, TFLOPs=66%2026.06 | 76.44 | |
| ShortVBackbone=Qwen3-VL-8B, TFLOPs=58%, N=202026.06 | 76.08 | |
| VSkipBackbone=Qwen3-VL-8B, TFLOPs=76%, N=202026.06 | 75.98 | |
| V2DropBackbone=Qwen2.5-VL-7B, TFLOPs=55%2026.06 | 72.26 | |
| Op-SkipBackbone=Qwen2.5-VL-7B, TFLOPs=55%, N=202026.06 | 72.02 | |
| ShortVBackbone=Qwen2.5-VL-7B, TFLOPs=50%, N=202026.06 | 71.93 | |
| APETBackbone=Qwen2.5-VL-7B, TFLOPs=55%2026.06 | 70.9 | |
| VTWBackbone=Qwen2.5-VL-7B, TFLOPs=50%, K=162026.06 | 65.43 | |
| VSkipBackbone=Qwen2.5-VL-7B, TFLOPs=67%, N=162026.06 | 63.58 | |
| VanillaBackbone=LLaVA-1.5-7B, TFLOPs=100%2026.06 | 58.21 | |
| VSkipBackbone=LLaVA-1.5-7B, TFLOPs=76%, N=202026.06 | 57.21 | |
| VSkip+Backbone=LLaVA-1.5-7B, TFLOPs=76%, N=202026.06 | 57.21 | |
| Op-SkipBackbone=LLaVA-1.5-7B, TFLOPs=66%, N=202026.06 | 56.4 | |
| APETBackbone=LLaVA-1.5-7B, TFLOPs=66%2026.06 | 56.36 | |
| V2DropBackbone=LLaVA-1.5-7B, TFLOPs=66%2026.06 | 55.92 | |
| ShortVBackbone=LLaVA-1.5-7B, TFLOPs=55%, N=202026.06 | 53.63 | |
| VTWBackbone=LLaVA-1.5-7B, TFLOPs=55%, K=162026.06 | 52.17 | |
| VTWBackbone=Qwen3-VL-8B, TFLOPs=58%, K=162026.06 | 51.16 |