Image-to-Web Conversion on IW-BENCH (Complex)
20.4Element AccuracyWebSight
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| WebSightChain-of-Thought (MCoT) status=Disabled2024.09 | 20.4 | 19.1 | |
| Qwen-VL-MaxChain-of-Thought (MCoT) status=Enabled, MCoT Hops=52024.09 | 19.5 | 18.3 | |
| Qwen-VL-MaxChain-of-Thought (MCoT) status=Disabled2024.09 | 17.6 | 16.4 | |
| GPT4VChain-of-Thought (MCoT) status=Enabled, MCoT Hops=52024.09 | 16.7 | 15.9 | |
| Gemini ProChain-of-Thought (MCoT) status=Enabled, MCoT Hops=52024.09 | 16.2 | 15.8 | |
| Gemini ProChain-of-Thought (MCoT) status=Disabled2024.09 | 15.3 | 14.5 | |
| Claude 3 OpusChain-of-Thought (MCoT) status=Enabled, MCoT Hops=52024.09 | 15.3 | 14.9 | |
| GPT4VChain-of-Thought (MCoT) status=Disabled2024.09 | 14.5 | 13.7 | |
| Claude 3 OpusChain-of-Thought (MCoT) status=Disabled2024.09 | 14.1 | 13.8 | |
| Qwen-VL-PlusChain-of-Thought (MCoT) status=Disabled2024.09 | 13.8 | 13.1 | |
| Qwen-VL-ChatChain-of-Thought (MCoT) status=Disabled2024.09 | 11.9 | 11.2 | |
| LLAVA-LLAMA-2-13BChain-of-Thought (MCoT) status=Enabled, MCoT Hops=52024.09 | 8.9 | 8.3 | |
| MPLUG-OWL2Chain-of-Thought (MCoT) status=Disabled2024.09 | 7.3 | 6 | |
| LLaVA-LLaMA-2-13BChain-of-Thought (MCoT) status=Disabled2024.09 | 6.3 | 5.9 | |
| miniGPT-4-LLaMA-2-7BChain-of-Thought (MCoT) status=Disabled2024.09 | 4.2 | 4 | |
| LLaMA-Adapter-V2-7BChain-of-Thought (MCoT) status=Disabled2024.09 | 4.1 | 4 |