Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

BridgeEQA

Benchmarks

Task NameDataset NameSOTA ResultTrend
Embodied Question AnsweringBridgeEQA 1,100 QA pairs (test)
Answer Correctness64.8
15
Embodied Question AnsweringBridgeEQA (test)
Image Citation Relevance88.9
15
Condition RatingBridgeEQA (instances with < 30 images)
Exact Match Accuracy40.9
9
Condition RatingBridgeEQA fewer than 30 images
Condition Rating Accuracy (±1)81.8
9
Showing 4 of 4 rows