COIN
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
COIN
97.8Accuracy
13
COIN (test)
54.1Top-1 Accuracy
13
COIN
67.3Top-1 Accuracy
12
COIN
90.4Top-1 Acc
12
COIN
90Accuracy
12
CoIN
67.4Average Performance
11
COIN (test)
66.4Top-1 Acc
11
CoIN Avg
82.36SciQA
10
COIN (test)
63.4Step Recognition Top-1 Acc
10
CoIN
77.84SciQA Score
9
COIN (test)
92.7Top-1 Acc
9
CoIN Last
79.72SciQA
8
COIN
59.6Accuracy
8
COIN T=5 (test)
16.06SR
8
CoIN
61.42CoIN Average
7
CoIN benchmark
83.59ScienceQA Accuracy
6
COIN CT (Gardening) (test)
25MR (Mean Rank)
6
COIN CT Woodworking (test)
30Mean Rank
6
COIN CT (Cooking) (test)
6MR
6
CoIN Unseen (QAsk-Nav annotations) (val)
9.63SR
6
CoIN Seen Synonyms (QAsk-Nav annotations) (val)
15.88SR
6
CoIN Seen QAsk-Nav annotations (val)
12.2Success Rate (SR)
6
COIN (test)
25.5SR (T=3)
6
COIN (test)
57.045CR (%)
6
COIN-SV (test)
0.572AUC
6