HyperAI
HomeNewsLatest PapersTutorialsDatasetsWikiSOTALLM ModelsGPU LeaderboardEvents
Search
About
HyperAI

Visual Reasoning On Bongard Openworld

Metrics

2-Class Accuracy

Results

Performance results of various models on this benchmark

Model Name
2-Class Accuracy
Paper TitleRepository
Human91.0Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World-
ChatCaptioner + ChatGPT49.3Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World-
Componential analysis - gpt-4o92.8Cognitive Paradigms for Evaluating VLMs on Visual Reasoning Task-
componential analysis - gemini-2.093.6Cognitive Paradigms for Evaluating VLMs on Visual Reasoning Task-
BLIP-2 + ChatGPT (Fine-tuned)63.3Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World-
InstructBLIP + ChatGPT + Neuro-Symbolic55.5Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World-
SNAIL64.0Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World-
Otter49.3Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World-
InstructBLIP + GPT-463.8Bongard-OpenWorld: Few-Shot Reasoning for Free-form Visual Concepts in the Real World-
0 of 9 row(s) selected.