Visual Question Answering on A-OKVQA Open-Ended
72.14AccuracyCVLM
Evaluation Results
| Method | Links | |
|---|---|---|
| CVLMLLM=Qwen-VL2024.02 | 72.14 | |
| CVLM (3M IKPairs, Objects=3)LLM=Vicuna-7B, Objects=32024.02 | 70.91 | |
| CVLM (3M IKPairs, Objects=1)LLM=Vicuna-7B, Objects=12024.02 | 70.48 | |
| CVLM (3M IKPairs, Objects=5)LLM=Vicuna-7B, Objects=52024.02 | 70.39 | |
| CVLM (3M IKPairs) w/o FKALLM=Vicuna-7B2024.02 | 70.22 | |
| CVLM (3M IKPairs, Objects=8)LLM=Vicuna-7B, Objects=82024.02 | 69.08 | |
| InstructBLIPLLM=Vicuna-7B2024.02 | 67.86 | |
| Qwen-VLLLM=Qwen2024.02 | 67.25 | |
| CVLMLLM=Vicuna-7B2024.02 | 66.99 | |
| CVLM w/o FKALLM=Vicuna-7B2024.02 | 65.59 | |
| LLaVA-v1.5†LLM=Vicuna-7B2024.02 | 65.27 | |
| CVLM w/o (FKA & VKA)LLM=Vicuna-7B2024.02 | 64.37 | |
| InstructBLIPLLM=Flan-T5-XXL2024.02 | 62.01 | |
| InstructBLIPLLM=Flan-T5-XL2024.02 | 61.05 | |
| BLIP2LLM=Flan-T5-XXL2024.02 | 53.71 |