Visual Reasoning On Bongard Openworld
المقاييس
2-Class Accuracy
النتائج
نتائج أداء النماذج المختلفة على هذا المعيار القياسي
| Paper Title | ||
|---|---|---|
| componential analysis - gemini-2.0 | 93.6 | A Cognitive Paradigm Approach to Probe the Perception-Reasoning Interface in VLMs |
| Componential analysis - gpt-4o | 92.8 | A Cognitive Paradigm Approach to Probe the Perception-Reasoning Interface in VLMs |
| Human | 91.0 | Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World |
| SNAIL | 64.0 | Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World |
| InstructBLIP + GPT-4 | 63.8 | Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World |
| BLIP-2 + ChatGPT (Fine-tuned) | 63.3 | Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World |
| InstructBLIP + ChatGPT + Neuro-Symbolic | 55.5 | Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World |
| ChatCaptioner + ChatGPT | 49.3 | Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World |
| Otter | 49.3 | Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World |
0 of 9 row(s) selected.