Web UI Localization on Screenspot V2
82.9Center AccuracyDAVE
Evaluation Results
| Method | Links | |
|---|---|---|
| DAVELLM Backbone=Qwen-2.5-7B-Instruct2025.12 | 82.9 | |
| SigLIP 2LLM Backbone=Qwen-2.5-7B-Instruct2025.12 | 81.4 | |
| AIMv2LLM Backbone=Qwen-2.5-7B-Instruct2025.12 | 79.3 | |
| DophineLLM Backbone=Qwen-2.5-7B-Instruct2025.12 | 75.2 | |
| MAE-WebLLM Backbone=Qwen-2.5-7B-Instruct2025.12 | 70.1 | |
| DAVELLM Backbone=Llama-3.2-3B-Instruct2025.12 | 64.5 | |
| DophineLLM Backbone=Llama-3.2-3B-Instruct2025.12 | 56.3 | |
| MAE-WebLLM Backbone=Llama-3.2-3B-Instruct2025.12 | 51.1 | |
| Pix2structLLM Backbone=Qwen-2.5-7B-Instruct2025.12 | 47.8 | |
| SigLIP 2LLM Backbone=Llama-3.2-3B-Instruct2025.12 | 40.7 | |
| Pix2StructLLM Backbone=Llama-3.2-3B-Instruct2025.12 | 32.6 | |
| AIMv2LLM Backbone=Llama-3.2-3B-Instruct2025.12 | 32 | |
| DinoV2LLM Backbone=Llama-3.2-3B-Instruct2025.12 | 16.5 | |
| DiTLLM Backbone=Llama-3.2-3B-Instruct2025.12 | 2.5 |