Dynamic Comprehension Evaluation on GUI-KRB
6.4Error Rate (%)GUI-explorer
Evaluation results
| Method | Links | |
|---|---|---|
| GUI-explorerKnowledge Ranking=true2025.05 | 6.4 | |
| GUI-explorer (w/o Ranker)Knowledge Ranking=false2025.05 | 6.8 | |
| Gemini 2.0 Flash2025.05 | 11.2 | |
| GPT-4o2025.05 | 13.4 | |
| Qwen2.5-VLBackbone=Qwen2.5-VL-7B-Instruct2025.05 | 14 | |
| Qwen2-VLBackbone=Qwen2-VL-72B-Instruct, Quantization=GPTQ-Int42025.05 | 19.8 |