Matches in SemOpenAlex for { <https://semopenalex.org/work/W3034407524> ?p ?o ?g. }
Showing items 1 to 88 of
88
with 100 items per page.
- W3034407524 abstract "Building general reading comprehension systems, capable of solving multiple datasets at the same time, is a recent aspirational goal in the research community. Prior work has focused on model architecture or generalization to held out datasets, and largely passed over the particulars of the multi-task learning set up. We show that a simple dynamic sampling strategy, selecting instances for training proportional to the multi-task model’s current performance on a dataset relative to its single task performance, gives substantive gains over prior multi-task sampling strategies, mitigating the catastrophic forgetting that is common in multi-task learning. We also demonstrate that allowing instances of different tasks to be interleaved as much as possible between each epoch and batch has a clear benefit in multitask performance over forcing task homogeneity at the epoch or batch level. Our final model shows greatly increased performance over the best model on ORB, a recently-released multitask reading comprehension benchmark." @default.
- W3034407524 created "2020-06-19" @default.
- W3034407524 creator A5005779128 @default.
- W3034407524 creator A5035088083 @default.
- W3034407524 creator A5044422816 @default.
- W3034407524 creator A5076846819 @default.
- W3034407524 date "2020-01-01" @default.
- W3034407524 modified "2023-10-16" @default.
- W3034407524 title "Dynamic Sampling Strategies for Multi-Task Reading Comprehension" @default.
- W3034407524 cites W2106304233 @default.
- W3034407524 cites W2557764419 @default.
- W3034407524 cites W2593765977 @default.
- W3034407524 cites W2626967530 @default.
- W3034407524 cites W2913340405 @default.
- W3034407524 cites W2919420119 @default.
- W3034407524 cites W2951036431 @default.
- W3034407524 cites W2951873305 @default.
- W3034407524 cites W2963168538 @default.
- W3034407524 cites W2963323070 @default.
- W3034407524 cites W2963748441 @default.
- W3034407524 cites W2963928014 @default.
- W3034407524 cites W2963963993 @default.
- W3034407524 cites W2970745243 @default.
- W3034407524 cites W2987553933 @default.
- W3034407524 cites W2988421999 @default.
- W3034407524 cites W2997924070 @default.
- W3034407524 doi "https://doi.org/10.18653/v1/2020.acl-main.86" @default.
- W3034407524 hasPublicationYear "2020" @default.
- W3034407524 type Work @default.
- W3034407524 sameAs 3034407524 @default.
- W3034407524 citedByCount "7" @default.
- W3034407524 countsByYear W30344075242021 @default.
- W3034407524 countsByYear W30344075242022 @default.
- W3034407524 crossrefType "proceedings-article" @default.
- W3034407524 hasAuthorship W3034407524A5005779128 @default.
- W3034407524 hasAuthorship W3034407524A5035088083 @default.
- W3034407524 hasAuthorship W3034407524A5044422816 @default.
- W3034407524 hasAuthorship W3034407524A5076846819 @default.
- W3034407524 hasBestOaLocation W30344075241 @default.
- W3034407524 hasConcept C119857082 @default.
- W3034407524 hasConcept C154945302 @default.
- W3034407524 hasConcept C15744967 @default.
- W3034407524 hasConcept C162324750 @default.
- W3034407524 hasConcept C177264268 @default.
- W3034407524 hasConcept C17744445 @default.
- W3034407524 hasConcept C180747234 @default.
- W3034407524 hasConcept C187736073 @default.
- W3034407524 hasConcept C199360897 @default.
- W3034407524 hasConcept C199539241 @default.
- W3034407524 hasConcept C2778780117 @default.
- W3034407524 hasConcept C2780451532 @default.
- W3034407524 hasConcept C28006648 @default.
- W3034407524 hasConcept C41008148 @default.
- W3034407524 hasConcept C554936623 @default.
- W3034407524 hasConcept C7149132 @default.
- W3034407524 hasConceptScore W3034407524C119857082 @default.
- W3034407524 hasConceptScore W3034407524C154945302 @default.
- W3034407524 hasConceptScore W3034407524C15744967 @default.
- W3034407524 hasConceptScore W3034407524C162324750 @default.
- W3034407524 hasConceptScore W3034407524C177264268 @default.
- W3034407524 hasConceptScore W3034407524C17744445 @default.
- W3034407524 hasConceptScore W3034407524C180747234 @default.
- W3034407524 hasConceptScore W3034407524C187736073 @default.
- W3034407524 hasConceptScore W3034407524C199360897 @default.
- W3034407524 hasConceptScore W3034407524C199539241 @default.
- W3034407524 hasConceptScore W3034407524C2778780117 @default.
- W3034407524 hasConceptScore W3034407524C2780451532 @default.
- W3034407524 hasConceptScore W3034407524C28006648 @default.
- W3034407524 hasConceptScore W3034407524C41008148 @default.
- W3034407524 hasConceptScore W3034407524C554936623 @default.
- W3034407524 hasConceptScore W3034407524C7149132 @default.
- W3034407524 hasLocation W30344075241 @default.
- W3034407524 hasOpenAccess W3034407524 @default.
- W3034407524 hasPrimaryLocation W30344075241 @default.
- W3034407524 hasRelatedWork W10944326 @default.
- W3034407524 hasRelatedWork W12016155 @default.
- W3034407524 hasRelatedWork W12824513 @default.
- W3034407524 hasRelatedWork W13538861 @default.
- W3034407524 hasRelatedWork W14777649 @default.
- W3034407524 hasRelatedWork W14789944 @default.
- W3034407524 hasRelatedWork W14828854 @default.
- W3034407524 hasRelatedWork W3569651 @default.
- W3034407524 hasRelatedWork W7339520 @default.
- W3034407524 hasRelatedWork W9761094 @default.
- W3034407524 isParatext "false" @default.
- W3034407524 isRetracted "false" @default.
- W3034407524 magId "3034407524" @default.
- W3034407524 workType "article" @default.