Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387223265> ?p ?o ?g. }
Showing items 1 to 47 of
47
with 100 items per page.
- W4387223265 abstract "We introduce Transformer-VQ, a decoder-only transformer computing softmax-based dense self-attention in linear time. Transformer-VQ's efficient attention is enabled by vector-quantized keys and a novel caching mechanism. In large-scale experiments, Transformer-VQ is shown highly competitive in quality, with strong results on Enwik8 (0.99 bpb), PG-19 (26.6 ppl), and ImageNet64 (3.16 bpb). Code: https://github.com/transformer-vq/transformer_vq" @default.
- W4387223265 created "2023-10-01" @default.
- W4387223265 creator A5078064866 @default.
- W4387223265 date "2023-09-28" @default.
- W4387223265 modified "2023-10-01" @default.
- W4387223265 title "Transformer-VQ: Linear-Time Transformers via Vector Quantization" @default.
- W4387223265 doi "https://doi.org/10.48550/arxiv.2309.16354" @default.
- W4387223265 hasPublicationYear "2023" @default.
- W4387223265 type Work @default.
- W4387223265 citedByCount "0" @default.
- W4387223265 crossrefType "posted-content" @default.
- W4387223265 hasAuthorship W4387223265A5078064866 @default.
- W4387223265 hasBestOaLocation W43872232651 @default.
- W4387223265 hasConcept C11413529 @default.
- W4387223265 hasConcept C119599485 @default.
- W4387223265 hasConcept C127413603 @default.
- W4387223265 hasConcept C165801399 @default.
- W4387223265 hasConcept C199833920 @default.
- W4387223265 hasConcept C28490314 @default.
- W4387223265 hasConcept C33923547 @default.
- W4387223265 hasConcept C41008148 @default.
- W4387223265 hasConcept C66322947 @default.
- W4387223265 hasConceptScore W4387223265C11413529 @default.
- W4387223265 hasConceptScore W4387223265C119599485 @default.
- W4387223265 hasConceptScore W4387223265C127413603 @default.
- W4387223265 hasConceptScore W4387223265C165801399 @default.
- W4387223265 hasConceptScore W4387223265C199833920 @default.
- W4387223265 hasConceptScore W4387223265C28490314 @default.
- W4387223265 hasConceptScore W4387223265C33923547 @default.
- W4387223265 hasConceptScore W4387223265C41008148 @default.
- W4387223265 hasConceptScore W4387223265C66322947 @default.
- W4387223265 hasLocation W43872232651 @default.
- W4387223265 hasOpenAccess W4387223265 @default.
- W4387223265 hasPrimaryLocation W43872232651 @default.
- W4387223265 hasRelatedWork W1992926600 @default.
- W4387223265 hasRelatedWork W2108450058 @default.
- W4387223265 hasRelatedWork W2110652681 @default.
- W4387223265 hasRelatedWork W2314954340 @default.
- W4387223265 hasRelatedWork W2368768466 @default.
- W4387223265 hasRelatedWork W2368779261 @default.
- W4387223265 hasRelatedWork W2778699561 @default.
- W4387223265 hasRelatedWork W3157910026 @default.
- W4387223265 hasRelatedWork W4246349004 @default.
- W4387223265 hasRelatedWork W55622562 @default.
- W4387223265 isParatext "false" @default.
- W4387223265 isRetracted "false" @default.
- W4387223265 workType "article" @default.