Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386076043> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W4386076043 abstract "We introduce À-la-carte Prompt Tuning (APT), a transformer-based scheme to tune prompts on distinct data so that they can be arbitrarily composed at inference time. The individual prompts can be trained in isolation, possibly on different devices, at different times, and on different distributions or domains. Furthermore each prompt only contains information about the subset of data it was exposed to during training. During inference, models can be assembled based on arbitrary selections of data sources, which we call à-la-carte learning. À-la-carte learning enables constructing bespoke models specific to each user's individual access rights and preferences. We can add or remove information from the model by simply adding or removing the corresponding prompts without retraining from scratch. We demonstrate that à-la-carte built models achieve accuracy within 5% of models trained on the union of the respective sources, with comparable cost in terms of training and inference time. For the continual learning benchmarks Split CIFAR- 100 and CORe50, we achieve state-of-the-art performance." @default.
- W4386076043 created "2023-08-23" @default.
- W4386076043 creator A5004410089 @default.
- W4386076043 creator A5005923362 @default.
- W4386076043 creator A5030073760 @default.
- W4386076043 creator A5034085630 @default.
- W4386076043 creator A5036782268 @default.
- W4386076043 creator A5038328783 @default.
- W4386076043 creator A5065386783 @default.
- W4386076043 date "2023-06-01" @default.
- W4386076043 modified "2023-10-18" @default.
- W4386076043 title "À-la-carte Prompt Tuning (APT): Combining Distinct Data Via Composable Prompting" @default.
- W4386076043 cites W1977295328 @default.
- W4386076043 cites W2138011018 @default.
- W4386076043 cites W2152161678 @default.
- W4386076043 cites W3035644192 @default.
- W4386076043 cites W3035682985 @default.
- W4386076043 cites W3095155273 @default.
- W4386076043 cites W3154155772 @default.
- W4386076043 cites W3174770825 @default.
- W4386076043 cites W3175430527 @default.
- W4386076043 cites W4205991051 @default.
- W4386076043 cites W4225484930 @default.
- W4386076043 cites W4285247752 @default.
- W4386076043 cites W4312238419 @default.
- W4386076043 cites W4312642299 @default.
- W4386076043 doi "https://doi.org/10.1109/cvpr52729.2023.01439" @default.
- W4386076043 hasPublicationYear "2023" @default.
- W4386076043 type Work @default.
- W4386076043 citedByCount "0" @default.
- W4386076043 crossrefType "proceedings-article" @default.
- W4386076043 hasAuthorship W4386076043A5004410089 @default.
- W4386076043 hasAuthorship W4386076043A5005923362 @default.
- W4386076043 hasAuthorship W4386076043A5030073760 @default.
- W4386076043 hasAuthorship W4386076043A5034085630 @default.
- W4386076043 hasAuthorship W4386076043A5036782268 @default.
- W4386076043 hasAuthorship W4386076043A5038328783 @default.
- W4386076043 hasAuthorship W4386076043A5065386783 @default.
- W4386076043 hasConcept C119599485 @default.
- W4386076043 hasConcept C119857082 @default.
- W4386076043 hasConcept C127413603 @default.
- W4386076043 hasConcept C134306372 @default.
- W4386076043 hasConcept C144133560 @default.
- W4386076043 hasConcept C154945302 @default.
- W4386076043 hasConcept C155202549 @default.
- W4386076043 hasConcept C165801399 @default.
- W4386076043 hasConcept C17744445 @default.
- W4386076043 hasConcept C199360897 @default.
- W4386076043 hasConcept C199539241 @default.
- W4386076043 hasConcept C2776214188 @default.
- W4386076043 hasConcept C2778712577 @default.
- W4386076043 hasConcept C2781235140 @default.
- W4386076043 hasConcept C33923547 @default.
- W4386076043 hasConcept C41008148 @default.
- W4386076043 hasConcept C44210515 @default.
- W4386076043 hasConcept C66322947 @default.
- W4386076043 hasConcept C77618280 @default.
- W4386076043 hasConceptScore W4386076043C119599485 @default.
- W4386076043 hasConceptScore W4386076043C119857082 @default.
- W4386076043 hasConceptScore W4386076043C127413603 @default.
- W4386076043 hasConceptScore W4386076043C134306372 @default.
- W4386076043 hasConceptScore W4386076043C144133560 @default.
- W4386076043 hasConceptScore W4386076043C154945302 @default.
- W4386076043 hasConceptScore W4386076043C155202549 @default.
- W4386076043 hasConceptScore W4386076043C165801399 @default.
- W4386076043 hasConceptScore W4386076043C17744445 @default.
- W4386076043 hasConceptScore W4386076043C199360897 @default.
- W4386076043 hasConceptScore W4386076043C199539241 @default.
- W4386076043 hasConceptScore W4386076043C2776214188 @default.
- W4386076043 hasConceptScore W4386076043C2778712577 @default.
- W4386076043 hasConceptScore W4386076043C2781235140 @default.
- W4386076043 hasConceptScore W4386076043C33923547 @default.
- W4386076043 hasConceptScore W4386076043C41008148 @default.
- W4386076043 hasConceptScore W4386076043C44210515 @default.
- W4386076043 hasConceptScore W4386076043C66322947 @default.
- W4386076043 hasConceptScore W4386076043C77618280 @default.
- W4386076043 hasLocation W43860760431 @default.
- W4386076043 hasOpenAccess W4386076043 @default.
- W4386076043 hasPrimaryLocation W43860760431 @default.
- W4386076043 hasRelatedWork W1842818334 @default.
- W4386076043 hasRelatedWork W2065109233 @default.
- W4386076043 hasRelatedWork W2961085424 @default.
- W4386076043 hasRelatedWork W3035617937 @default.
- W4386076043 hasRelatedWork W4288725058 @default.
- W4386076043 hasRelatedWork W4306887858 @default.
- W4386076043 hasRelatedWork W4312845548 @default.
- W4386076043 hasRelatedWork W4366250329 @default.
- W4386076043 hasRelatedWork W4385728911 @default.
- W4386076043 hasRelatedWork W4386076043 @default.
- W4386076043 isParatext "false" @default.
- W4386076043 isRetracted "false" @default.
- W4386076043 workType "article" @default.