Matches in SemOpenAlex for { <https://semopenalex.org/work/W2993313557> ?p ?o ?g. }
- W2993313557 abstract "Much of vision-and-language research focuses on a small but diverse set of independent tasks and supporting datasets often studied in isolation; however, the visually-grounded language understanding skills required for success at these tasks overlap significantly. In this work, we investigate these relationships between vision-and-language tasks by developing a large-scale, multi-task training regime. Our approach culminates in a single model on 12 datasets from four broad categories of task including visual question answering, caption-based image retrieval, grounding referring expressions, and multi-modal verification. Compared to independently trained single-task models, this represents a reduction from approximately 3 billion parameters to 270 million while simultaneously improving performance by 2.05 points on average across tasks. We use our multi-task framework to perform in-depth analysis of the effect of joint training diverse tasks. Further, we show that finetuning task-specific models from our single multi-task model can lead to further improvements, achieving performance at or above the state-of-the-art." @default.
- W2993313557 created "2019-12-13" @default.
- W2993313557 creator A5024481540 @default.
- W2993313557 creator A5035752789 @default.
- W2993313557 creator A5050342343 @default.
- W2993313557 creator A5051259505 @default.
- W2993313557 creator A5090130929 @default.
- W2993313557 date "2019-12-04" @default.
- W2993313557 modified "2023-10-16" @default.
- W2993313557 title "12-in-1: Multi-Task Vision and Language Representation Learning" @default.
- W2993313557 cites W1773149199 @default.
- W2993313557 cites W1889081078 @default.
- W2993313557 cites W1896424170 @default.
- W2993313557 cites W1996430422 @default.
- W2993313557 cites W2102674365 @default.
- W2993313557 cites W2109586012 @default.
- W2993313557 cites W2117130368 @default.
- W2993313557 cites W2174786457 @default.
- W2993313557 cites W2251512949 @default.
- W2993313557 cites W2277195237 @default.
- W2993313557 cites W2296073425 @default.
- W2993313557 cites W2558535589 @default.
- W2993313557 cites W2558809543 @default.
- W2993313557 cites W2560730294 @default.
- W2993313557 cites W2624871570 @default.
- W2993313557 cites W2745461083 @default.
- W2993313557 cites W2774005037 @default.
- W2993313557 cites W2795151422 @default.
- W2993313557 cites W2809324505 @default.
- W2993313557 cites W2886641317 @default.
- W2993313557 cites W2901479108 @default.
- W2993313557 cites W2914120296 @default.
- W2993313557 cites W2914526845 @default.
- W2993313557 cites W2914746235 @default.
- W2993313557 cites W2938082352 @default.
- W2993313557 cites W2946233749 @default.
- W2993313557 cites W2950104027 @default.
- W2993313557 cites W2950541952 @default.
- W2993313557 cites W2950813464 @default.
- W2993313557 cites W2950872548 @default.
- W2993313557 cites W2952603081 @default.
- W2993313557 cites W2953106684 @default.
- W2993313557 cites W2961117861 @default.
- W2993313557 cites W2962753370 @default.
- W2993313557 cites W2962964995 @default.
- W2993313557 cites W2963109634 @default.
- W2993313557 cites W2963115613 @default.
- W2993313557 cites W2963199420 @default.
- W2993313557 cites W2963341956 @default.
- W2993313557 cites W2963403868 @default.
- W2993313557 cites W2963498646 @default.
- W2993313557 cites W2963530300 @default.
- W2993313557 cites W2963540523 @default.
- W2993313557 cites W2963668159 @default.
- W2993313557 cites W2963800628 @default.
- W2993313557 cites W2963877604 @default.
- W2993313557 cites W2964345792 @default.
- W2993313557 cites W2965373594 @default.
- W2993313557 cites W2967593235 @default.
- W2993313557 cites W2968124245 @default.
- W2993313557 cites W2968880719 @default.
- W2993313557 cites W2970231061 @default.
- W2993313557 cites W2970608575 @default.
- W2993313557 cites W2975501350 @default.
- W2993313557 cites W2981468122 @default.
- W2993313557 cites W2981852735 @default.
- W2993313557 cites W2982152811 @default.
- W2993313557 cites W2995460200 @default.
- W2993313557 cites W2997591391 @default.
- W2993313557 cites W3020257313 @default.
- W2993313557 cites W3034337319 @default.
- W2993313557 cites W3037108562 @default.
- W2993313557 doi "https://doi.org/10.48550/arxiv.1912.02315" @default.
- W2993313557 hasPublicationYear "2019" @default.
- W2993313557 type Work @default.
- W2993313557 sameAs 2993313557 @default.
- W2993313557 citedByCount "8" @default.
- W2993313557 countsByYear W29933135572020 @default.
- W2993313557 countsByYear W29933135572021 @default.
- W2993313557 countsByYear W29933135572022 @default.
- W2993313557 crossrefType "posted-content" @default.
- W2993313557 hasAuthorship W2993313557A5024481540 @default.
- W2993313557 hasAuthorship W2993313557A5035752789 @default.
- W2993313557 hasAuthorship W2993313557A5050342343 @default.
- W2993313557 hasAuthorship W2993313557A5051259505 @default.
- W2993313557 hasAuthorship W2993313557A5090130929 @default.
- W2993313557 hasBestOaLocation W29933135571 @default.
- W2993313557 hasConcept C107457646 @default.
- W2993313557 hasConcept C119857082 @default.
- W2993313557 hasConcept C121332964 @default.
- W2993313557 hasConcept C137293760 @default.
- W2993313557 hasConcept C154945302 @default.
- W2993313557 hasConcept C162324750 @default.
- W2993313557 hasConcept C175154964 @default.
- W2993313557 hasConcept C177264268 @default.
- W2993313557 hasConcept C17744445 @default.
- W2993313557 hasConcept C185592680 @default.
- W2993313557 hasConcept C187736073 @default.
- W2993313557 hasConcept C188027245 @default.
- W2993313557 hasConcept C199360897 @default.