Visual Question Answering on GQA (GQA Score)
62.3GQA ScoreSemHiTok-384
Evaluation Results
| Method | Links | |
|---|---|---|
| SemHiTok-384LLM=Vicuna-7B, Data=LLaVA-v1.5, Resolution=3842025.03 | 62.3 | |
| SigLIPLLM=Vicuna-7B, Data=LLaVA-v1.5, Resolution=2562025.03 | 61.9 | |
| TokenFlow-384LLM=Vicuna-7B, Data=LLaVA-v1.5, Resolution=3842025.03 | 61.2 | |
| SemHiTokLLM=Vicuna-7B, Data=LLaVA-v1.5, Resolution=2562025.03 | 60.3 | |
| TokLIPLLM=Qwen2.5-7B-Ins, Data=LLaVA-v1.5, Resolution=3842025.03 | 59.3 | |
| SDE(MUSE-VL)LLM=Vicuna-7B, Data=LLaVA-v1.5, Resolution=2562025.03 | 58 | |
| TokLIPLLM=Qwen2.5-7B-Ins, Data=LLaVA-v1.5, Resolution=2562025.03 | 57.4 | |
| VILA-ULLM=Vicuna-7B, Data=LLaVA-v1.5, Resolution=2562025.03 | 55.3 | |
| LlamaGenLLM=Vicuna-7B, Data=LLaVA-v1.5, Resolution=2562025.03 | 39.8 |