HyperAI초신경

Image Sentence Alignment On Valse Existence

평가 지표

Accuracy (%)

pairwise accuracy

평가 결과

이 벤치마크에서 각 모델의 성능 결과

			Paper Title
ViLBERT 12-in-1	89.0	95.6	VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena
LXMERT	55.8	78.6	VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena
CLIP	-	66.9	VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena
ViLBERT	2.4	66.5	VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena
GPT1	-	61.8	VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena
GPT2	-	58.0	VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena
VisualBERT	49.3	39.7	VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena

0 of 7 row(s) selected.

Image Sentence Alignment On Valse Existence | SOTA | HyperAI초신경