Matches in SemOpenAlex for { <https://semopenalex.org/work/W2912018747> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W2912018747 abstract "Adaptive gradient-based optimizers such as Adagrad and Adam are crucial for achieving state-of-the-art performance in machine translation and language modeling. However, these methods maintain second-order statistics for each parameter, thus introducing significant memory overheads that restrict the size of the model being used as well as the number of examples in a mini-batch. We describe an effective and flexible adaptive optimization method with greatly reduced memory overhead. Our method retains the benefits of per-parameter adaptivity while allowing significantly larger models and batch sizes. We give convergence guarantees for our method, and demonstrate its effectiveness in training very large translation and language models with up to 2-fold speedups compared to the state-of-the-art." @default.
- W2912018747 created "2019-02-21" @default.
- W2912018747 creator A5038867895 @default.
- W2912018747 creator A5050591232 @default.
- W2912018747 creator A5053775980 @default.
- W2912018747 creator A5057336278 @default.
- W2912018747 date "2019-01-30" @default.
- W2912018747 modified "2023-09-27" @default.
- W2912018747 title "Memory-Efficient Adaptive Optimization for Large-Scale Learning." @default.
- W2912018747 hasPublicationYear "2019" @default.
- W2912018747 type Work @default.
- W2912018747 sameAs 2912018747 @default.
- W2912018747 citedByCount "9" @default.
- W2912018747 countsByYear W29120187472019 @default.
- W2912018747 countsByYear W29120187472020 @default.
- W2912018747 countsByYear W29120187472021 @default.
- W2912018747 crossrefType "posted-content" @default.
- W2912018747 hasAuthorship W2912018747A5038867895 @default.
- W2912018747 hasAuthorship W2912018747A5050591232 @default.
- W2912018747 hasAuthorship W2912018747A5053775980 @default.
- W2912018747 hasAuthorship W2912018747A5057336278 @default.
- W2912018747 hasConcept C104317684 @default.
- W2912018747 hasConcept C105580179 @default.
- W2912018747 hasConcept C11413529 @default.
- W2912018747 hasConcept C121332964 @default.
- W2912018747 hasConcept C126255220 @default.
- W2912018747 hasConcept C137293760 @default.
- W2912018747 hasConcept C149364088 @default.
- W2912018747 hasConcept C149672232 @default.
- W2912018747 hasConcept C154945302 @default.
- W2912018747 hasConcept C162324750 @default.
- W2912018747 hasConcept C185592680 @default.
- W2912018747 hasConcept C199360897 @default.
- W2912018747 hasConcept C203005215 @default.
- W2912018747 hasConcept C2777303404 @default.
- W2912018747 hasConcept C2778755073 @default.
- W2912018747 hasConcept C2779960059 @default.
- W2912018747 hasConcept C33923547 @default.
- W2912018747 hasConcept C41008148 @default.
- W2912018747 hasConcept C48103436 @default.
- W2912018747 hasConcept C50522688 @default.
- W2912018747 hasConcept C55493867 @default.
- W2912018747 hasConcept C62520636 @default.
- W2912018747 hasConceptScore W2912018747C104317684 @default.
- W2912018747 hasConceptScore W2912018747C105580179 @default.
- W2912018747 hasConceptScore W2912018747C11413529 @default.
- W2912018747 hasConceptScore W2912018747C121332964 @default.
- W2912018747 hasConceptScore W2912018747C126255220 @default.
- W2912018747 hasConceptScore W2912018747C137293760 @default.
- W2912018747 hasConceptScore W2912018747C149364088 @default.
- W2912018747 hasConceptScore W2912018747C149672232 @default.
- W2912018747 hasConceptScore W2912018747C154945302 @default.
- W2912018747 hasConceptScore W2912018747C162324750 @default.
- W2912018747 hasConceptScore W2912018747C185592680 @default.
- W2912018747 hasConceptScore W2912018747C199360897 @default.
- W2912018747 hasConceptScore W2912018747C203005215 @default.
- W2912018747 hasConceptScore W2912018747C2777303404 @default.
- W2912018747 hasConceptScore W2912018747C2778755073 @default.
- W2912018747 hasConceptScore W2912018747C2779960059 @default.
- W2912018747 hasConceptScore W2912018747C33923547 @default.
- W2912018747 hasConceptScore W2912018747C41008148 @default.
- W2912018747 hasConceptScore W2912018747C48103436 @default.
- W2912018747 hasConceptScore W2912018747C50522688 @default.
- W2912018747 hasConceptScore W2912018747C55493867 @default.
- W2912018747 hasConceptScore W2912018747C62520636 @default.
- W2912018747 hasLocation W29120187471 @default.
- W2912018747 hasOpenAccess W2912018747 @default.
- W2912018747 hasPrimaryLocation W29120187471 @default.
- W2912018747 hasRelatedWork W128831403 @default.
- W2912018747 hasRelatedWork W1857353606 @default.
- W2912018747 hasRelatedWork W1969805974 @default.
- W2912018747 hasRelatedWork W2005294557 @default.
- W2912018747 hasRelatedWork W2797328513 @default.
- W2912018747 hasRelatedWork W2895139741 @default.
- W2912018747 hasRelatedWork W2913618604 @default.
- W2912018747 hasRelatedWork W2947983473 @default.
- W2912018747 hasRelatedWork W2963341956 @default.
- W2912018747 hasRelatedWork W2963403868 @default.
- W2912018747 hasRelatedWork W2970294425 @default.
- W2912018747 hasRelatedWork W2972504565 @default.
- W2912018747 hasRelatedWork W2989832790 @default.
- W2912018747 hasRelatedWork W2995727426 @default.
- W2912018747 hasRelatedWork W3011763294 @default.
- W2912018747 hasRelatedWork W3013134448 @default.
- W2912018747 hasRelatedWork W3041516094 @default.
- W2912018747 hasRelatedWork W3089416163 @default.
- W2912018747 hasRelatedWork W3122102896 @default.
- W2912018747 hasRelatedWork W3167203501 @default.
- W2912018747 isParatext "false" @default.
- W2912018747 isRetracted "false" @default.
- W2912018747 magId "2912018747" @default.
- W2912018747 workType "article" @default.