paper-with-me

image-sentence alignment 벤치마크

image-sentence alignment on VALSE spatial relations

14개 결과 · ⬇ CSV · JSON

pairwise accuracy

39.7 49.08 58.45 67.83 77.2 2021-12 2026-09 GPT1 — 77.2 (2021-12-14) GPT2 — 75.0 (2021-12-14) ViLBERT 12-in-1 — 67.7 (2021-12-14) CLIP — 64.3 (2021-12-14) LXMERT — 60.2 (2021-12-14) ViLBERT — 57.2 (2021-12-14) VisualBERT — 39.7 (2021-12-14) GPT1 — 77.2 (2021-12-14) GPT2 — 75.0 (2021-12-14) ViLBERT 12-in-1 — 67.7 (2021-12-14) CLIP — 64.3 (2021-12-14) LXMERT — 60.2 (2021-12-14) ViLBERT — 57.2 (2021-12-14) VisualBERT — 39.7 (2021-12-14) GPT1 — 77.2 (2021-12-14)
RankModel pairwise accuracyAccuracy (%) PaperCodeYear
11 CLIP 64.3– VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena heidelberg-nlp/valse 2021
12 LXMERT 60.250.8 VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena heidelberg-nlp/valse 2021
13 ViLBERT 57.249.9 VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena heidelberg-nlp/valse 2021
14 VisualBERT 39.749.3 VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena heidelberg-nlp/valse 2021
← 이전 11–14 / 14 페이지당 10 20 50 100