COCO
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
COCO 10K prompts 2014 (Karpathy)
31.81HPS
6
COCO
29.81PSNR
6
COCO
38.74PSNR
6
COCO long prompts (EE attack)
4ASR (CLIP)
6
COCO long prompts TI attack
40ASR (CLIP)
6
COCO long prompts (RR attack)
71ASR (CLIP)
6
COCO Captioning (val)
5.6CHAIRs
6
COCO 2017 (val)
4,633Throughput (tok/s)
6
COCO FuzzingRL probes (held-out)
100Accuracy
6
COCO obj-80
41.9AP
6
COCOCap
63.7Score
6
COCO captions 2017 (test)
5.58Latency (s)
6
COCO 2014 (val)
13.5R@1
6
COCO
50.2AP
6
COCO Localized Narratives (test)
0.665BLEU-1
6
COCO all categories 80 classes
27.7AR10
6
COCO (test)
83F1 Score
6
COCO-QLTI
83.4R@10
6
COCO (test)
43.7AP
6
COCO 40-10 setting (val)
51.79PQ (Co)
6
COCO 40-40 setting (val)
51.79PQ (Things)
6
COCO
46.08mAP
6
COCO-10K
16.46FID
6
COCO-Object
76pAcc
6
RC COCO
28.15CropCLIP
6