Improved SNMN

Reported on 7 benchmarks across 1 task

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing7 results

Visual Question Answering (VQA)onGQA Test2019
Accuracy
56.09
best: 89.3 (human)
Visual Question Answering (VQA)onGQA Test2019
Binary
73.4
best: 91.2 (human)
Visual Question Answering (VQA)onGQA Test2019
Consistency
85.11
best: 98.4 (human)
Visual Question Answering (VQA)onGQA Test2019
Distribution
5.14
best: 93.08 (GlobalPrior)
Visual Question Answering (VQA)onGQA Test2019
Open
40.82
best: 87.4 (human)
Visual Question Answering (VQA)onGQA Test2019
Plausibility
84.79
best: 97.2 (human)
Visual Question Answering (VQA)onGQA Test2019
Validity
96.37
best: 98.9 (human)