Matches in SemOpenAlex for { <https://semopenalex.org/work/W2992304145> ?p ?o ?g. }
- W2992304145 abstract "Prior work in visual dialog has focused on training deep neural models on VisDial in isolation. Instead, we present an approach to leverage pretraining on related vision-language datasets before transferring to visual dialog. We adapt the recently proposed ViLBERT (Lu et al., 2019) model for multi-turn visually-grounded conversations. Our model is pretrained on the Conceptual Captions and Visual Question Answering datasets, and finetuned on VisDial. Our best single model outperforms prior published work (including model ensembles) by more than 1% absolute on NDCG and MRR. Next, we find that additional finetuning using annotations in VisDial leads to even higher NDCG -- more than 10% over our base model -- but hurts MRR -- more than 17% below our base model! This highlights a trade-off between the two primary metrics -- NDCG and MRR -- which we find is due to dense annotations not correlating well with the original ground-truth answers to questions." @default.
- W2992304145 created "2019-12-13" @default.
- W2992304145 creator A5014035752 @default.
- W2992304145 creator A5042265238 @default.
- W2992304145 creator A5049841785 @default.
- W2992304145 creator A5050342343 @default.
- W2992304145 date "2019-12-05" @default.
- W2992304145 modified "2023-10-16" @default.
- W2992304145 title "Large-scale Pretraining for Visual Dialog: A Simple State-of-the-Art Baseline" @default.
- W2992304145 cites W1566289585 @default.
- W2992304145 cites W1861492603 @default.
- W2992304145 cites W1933349210 @default.
- W2992304145 cites W2117539524 @default.
- W2992304145 cites W2163605009 @default.
- W2992304145 cites W2185175083 @default.
- W2992304145 cites W2194775991 @default.
- W2992304145 cites W2251512949 @default.
- W2992304145 cites W2277195237 @default.
- W2992304145 cites W2558809543 @default.
- W2992304145 cites W2599940792 @default.
- W2992304145 cites W2603266952 @default.
- W2992304145 cites W2613718673 @default.
- W2992304145 cites W2745461083 @default.
- W2992304145 cites W2768661419 @default.
- W2992304145 cites W2835434549 @default.
- W2992304145 cites W2886641317 @default.
- W2992304145 cites W2892245540 @default.
- W2992304145 cites W2912371042 @default.
- W2992304145 cites W2914204778 @default.
- W2992304145 cites W2917061951 @default.
- W2992304145 cites W2945087694 @default.
- W2992304145 cites W2962835968 @default.
- W2992304145 cites W2962964995 @default.
- W2992304145 cites W2963115613 @default.
- W2992304145 cites W2963187678 @default.
- W2992304145 cites W2963287297 @default.
- W2992304145 cites W2963310665 @default.
- W2992304145 cites W2963341956 @default.
- W2992304145 cites W2963403868 @default.
- W2992304145 cites W2963530300 @default.
- W2992304145 cites W2963623904 @default.
- W2992304145 cites W2963643760 @default.
- W2992304145 cites W2963791035 @default.
- W2992304145 cites W2964121744 @default.
- W2992304145 cites W2964213933 @default.
- W2992304145 cites W2964218959 @default.
- W2992304145 cites W2965373594 @default.
- W2992304145 cites W2967045987 @default.
- W2992304145 cites W2968124245 @default.
- W2992304145 cites W2970231061 @default.
- W2992304145 cites W2970340522 @default.
- W2992304145 cites W2970355596 @default.
- W2992304145 cites W2970597249 @default.
- W2992304145 cites W2970608575 @default.
- W2992304145 cites W2970971581 @default.
- W2992304145 cites W2975501350 @default.
- W2992304145 cites W2981851019 @default.
- W2992304145 cites W2981902456 @default.
- W2992304145 cites W2988617410 @default.
- W2992304145 cites W2995460200 @default.
- W2992304145 cites W2996428491 @default.
- W2992304145 cites W2997547717 @default.
- W2992304145 cites W2998356391 @default.
- W2992304145 cites W3029418112 @default.
- W2992304145 cites W3035103424 @default.
- W2992304145 cites W3082274269 @default.
- W2992304145 cites W2940092135 @default.
- W2992304145 hasPublicationYear "2019" @default.
- W2992304145 type Work @default.
- W2992304145 sameAs 2992304145 @default.
- W2992304145 citedByCount "3" @default.
- W2992304145 countsByYear W29923041452020 @default.
- W2992304145 countsByYear W29923041452021 @default.
- W2992304145 crossrefType "posted-content" @default.
- W2992304145 hasAuthorship W2992304145A5014035752 @default.
- W2992304145 hasAuthorship W2992304145A5042265238 @default.
- W2992304145 hasAuthorship W2992304145A5049841785 @default.
- W2992304145 hasAuthorship W2992304145A5050342343 @default.
- W2992304145 hasConcept C111368507 @default.
- W2992304145 hasConcept C119857082 @default.
- W2992304145 hasConcept C12725497 @default.
- W2992304145 hasConcept C127313418 @default.
- W2992304145 hasConcept C136764020 @default.
- W2992304145 hasConcept C146849305 @default.
- W2992304145 hasConcept C153083717 @default.
- W2992304145 hasConcept C154945302 @default.
- W2992304145 hasConcept C165696696 @default.
- W2992304145 hasConcept C173853756 @default.
- W2992304145 hasConcept C189430467 @default.
- W2992304145 hasConcept C204321447 @default.
- W2992304145 hasConcept C2777508537 @default.
- W2992304145 hasConcept C38652104 @default.
- W2992304145 hasConcept C41008148 @default.
- W2992304145 hasConcept C86037889 @default.
- W2992304145 hasConceptScore W2992304145C111368507 @default.
- W2992304145 hasConceptScore W2992304145C119857082 @default.
- W2992304145 hasConceptScore W2992304145C12725497 @default.
- W2992304145 hasConceptScore W2992304145C127313418 @default.
- W2992304145 hasConceptScore W2992304145C136764020 @default.
- W2992304145 hasConceptScore W2992304145C146849305 @default.