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

CHAIR

Benchmarks

Task NameDataset NameSOTA ResultTrend
Hallucination EvaluationCHAIR
CHAIR_s72.8
393
Object Hallucination EvaluationCHAIR
CHAIRi Score57
174
Caption Hallucination EvaluationCHAIR
CS Score58.8
122
Fine-Grained Sketch-Based Image Retrieval (FG-SBIR)Chair V2 (test)
Top-1 Accuracy89.69
72
Image CaptioningCHAIR
CHAIR_S69.8
71
Hallucination EvaluationCHAIR MSCOCO
CHAIR_S59.4
64
Object Hallucination EvaluationCHAIR MSCOCO v1.0 (val)
CHAIRs54.6
51
Object Hallucination in Open-ended CaptioningCHAIR (test)
CHAIR_S62.3
50
Hallucination EvaluationCHAIR MSCOCO 2014 (val)
CHAIRi26.2
45
Object Hallucination EvaluationCHAIR MSCOCO
CS Score62
42
Long-form generation hallucination evaluationCHAIR
CS Score58.8
36
Hallucination EvaluationCHAIR MSCOCO 2014
CHAIRs Score51.3
28
Hallucination MitigationCHAIR
CHAIR_S75
24
Object Hallucination AssessmentCHAIR (test)
CS (%)58.2
22
Object Hallucination MitigationCHAIR
CHAIRs Score69.8
22
Image CaptioningCHAIR (test)
Cs Score52.6
22
Visual HallucinationCHAIR
CHAIR Score15.3
21
Free-form Captioning Hallucination EvaluationCHAIR
CHAIR-S Score59.6
20
Hallucination EvaluationCHAIR (test)
CS Score50.9
20
Object Hallucination EvaluationCHAIR MS COCO based (test)
CHAIRs56.2
18
Hallucination EvaluationCHAIR (val)
CHAIRs59.5
16
Language Quality EvaluationCHAIR benchmark (test)
BLEU-119.2
16
Object Hallucination EvaluationCHAIR (val)
CHAIRs Score58.8
15
Caption Hallucination AssessmentCHAIR Zoom Blur (val)
CHAIRS Score59.8
14
Hallucination AssessmentCHAIR
CS52.8
12
Showing 25 of 38 rows