Matches in SemOpenAlex for { <https://semopenalex.org/work/W2949417870> ?p ?o ?g. }
- W2949417870 abstract "Although Transformer has achieved great successes on many NLP tasks, its heavy structure with fully-connected attention connections leads to dependencies on large training data. In this paper, we present Star-Transformer, a lightweight alternative by careful sparsification. To reduce model complexity, we replace the fully-connected structure with a star-shaped topology, in which every two non-adjacent nodes are connected through a shared relay node. Thus, complexity is reduced from quadratic to linear, while preserving capacity to capture both local composition and long-range dependency. The experiments on four tasks (22 datasets) show that Star-Transformer achieved significant improvements against the standard Transformer for the modestly sized datasets." @default.
- W2949417870 created "2019-06-27" @default.
- W2949417870 creator A5003418019 @default.
- W2949417870 creator A5012259826 @default.
- W2949417870 creator A5022934107 @default.
- W2949417870 creator A5044665993 @default.
- W2949417870 creator A5061237050 @default.
- W2949417870 creator A5088888083 @default.
- W2949417870 date "2019-02-25" @default.
- W2949417870 modified "2023-09-23" @default.
- W2949417870 title "Star-Transformer" @default.
- W2949417870 cites W1632114991 @default.
- W2949417870 cites W1832693441 @default.
- W2949417870 cites W1840435438 @default.
- W2949417870 cites W1879966306 @default.
- W2949417870 cites W1899794420 @default.
- W2949417870 cites W2120615054 @default.
- W2949417870 cites W2144578941 @default.
- W2949417870 cites W2157331557 @default.
- W2949417870 cites W2158899491 @default.
- W2949417870 cites W2250539671 @default.
- W2949417870 cites W2251939518 @default.
- W2949417870 cites W2267186426 @default.
- W2949417870 cites W2308720496 @default.
- W2949417870 cites W2470673105 @default.
- W2949417870 cites W2471349142 @default.
- W2949417870 cites W2556468274 @default.
- W2949417870 cites W2606780347 @default.
- W2949417870 cites W2962685628 @default.
- W2949417870 cites W2962739339 @default.
- W2949417870 cites W2962897020 @default.
- W2949417870 cites W2962902328 @default.
- W2949417870 cites W2963241825 @default.
- W2949417870 cites W2963341956 @default.
- W2949417870 cites W2963355447 @default.
- W2949417870 cites W2963403868 @default.
- W2949417870 cites W2963580443 @default.
- W2949417870 cites W2963625095 @default.
- W2949417870 cites W2963918774 @default.
- W2949417870 cites W2964121744 @default.
- W2949417870 cites W2964189376 @default.
- W2949417870 hasPublicationYear "2019" @default.
- W2949417870 type Work @default.
- W2949417870 sameAs 2949417870 @default.
- W2949417870 citedByCount "0" @default.
- W2949417870 crossrefType "posted-content" @default.
- W2949417870 hasAuthorship W2949417870A5003418019 @default.
- W2949417870 hasAuthorship W2949417870A5012259826 @default.
- W2949417870 hasAuthorship W2949417870A5022934107 @default.
- W2949417870 hasAuthorship W2949417870A5044665993 @default.
- W2949417870 hasAuthorship W2949417870A5061237050 @default.
- W2949417870 hasAuthorship W2949417870A5088888083 @default.
- W2949417870 hasConcept C11413529 @default.
- W2949417870 hasConcept C119599485 @default.
- W2949417870 hasConcept C121332964 @default.
- W2949417870 hasConcept C122306262 @default.
- W2949417870 hasConcept C127413603 @default.
- W2949417870 hasConcept C129844170 @default.
- W2949417870 hasConcept C163258240 @default.
- W2949417870 hasConcept C165801399 @default.
- W2949417870 hasConcept C184720557 @default.
- W2949417870 hasConcept C199845137 @default.
- W2949417870 hasConcept C2524010 @default.
- W2949417870 hasConcept C2778156585 @default.
- W2949417870 hasConcept C31258907 @default.
- W2949417870 hasConcept C33923547 @default.
- W2949417870 hasConcept C41008148 @default.
- W2949417870 hasConcept C62520636 @default.
- W2949417870 hasConcept C66322947 @default.
- W2949417870 hasConcept C71976206 @default.
- W2949417870 hasConceptScore W2949417870C11413529 @default.
- W2949417870 hasConceptScore W2949417870C119599485 @default.
- W2949417870 hasConceptScore W2949417870C121332964 @default.
- W2949417870 hasConceptScore W2949417870C122306262 @default.
- W2949417870 hasConceptScore W2949417870C127413603 @default.
- W2949417870 hasConceptScore W2949417870C129844170 @default.
- W2949417870 hasConceptScore W2949417870C163258240 @default.
- W2949417870 hasConceptScore W2949417870C165801399 @default.
- W2949417870 hasConceptScore W2949417870C184720557 @default.
- W2949417870 hasConceptScore W2949417870C199845137 @default.
- W2949417870 hasConceptScore W2949417870C2524010 @default.
- W2949417870 hasConceptScore W2949417870C2778156585 @default.
- W2949417870 hasConceptScore W2949417870C31258907 @default.
- W2949417870 hasConceptScore W2949417870C33923547 @default.
- W2949417870 hasConceptScore W2949417870C41008148 @default.
- W2949417870 hasConceptScore W2949417870C62520636 @default.
- W2949417870 hasConceptScore W2949417870C66322947 @default.
- W2949417870 hasConceptScore W2949417870C71976206 @default.
- W2949417870 hasLocation W29494178701 @default.
- W2949417870 hasOpenAccess W2949417870 @default.
- W2949417870 hasPrimaryLocation W29494178701 @default.
- W2949417870 hasRelatedWork W2901028208 @default.
- W2949417870 hasRelatedWork W2915716523 @default.
- W2949417870 hasRelatedWork W2954731415 @default.
- W2949417870 hasRelatedWork W2956480774 @default.
- W2949417870 hasRelatedWork W3008851394 @default.
- W2949417870 hasRelatedWork W3033529678 @default.
- W2949417870 hasRelatedWork W3114896399 @default.
- W2949417870 hasRelatedWork W3115860342 @default.
- W2949417870 hasRelatedWork W3119866685 @default.