Matches in SemOpenAlex for { <https://semopenalex.org/work/W3102403234> ?p ?o ?g. }
Showing items 1 to 76 of
76
with 100 items per page.
- W3102403234 endingPage "5752" @default.
- W3102403234 startingPage "5741" @default.
- W3102403234 abstract "We introduce Bayesian Bits, a practical method for joint mixed precision quantization and pruning through gradient based optimization. Bayesian Bits employs a novel decomposition of the quantization operation, which sequentially considers doubling the bit width. At each new bit width, the residual error between the full precision value and the previously rounded value is quantized. We then decide whether or not to add this quantized residual error for a higher effective bit width and lower quantization noise. By starting with a power-of-two bit width, this decomposition will always produce hardware-friendly configurations, and through an additional 0-bit option, serves as a unified view of pruning and quantization. Bayesian Bits then introduces learnable stochastic gates, which collectively control the bit width of the given tensor. As a result, we can obtain low bit solutions by performing approximate inference over the gates, with prior distributions that encourage most of them to be switched off. We experimentally validate our proposed method on several benchmark datasets and show that we can learn pruned, mixed precision networks that provide a better trade-off between accuracy and efficiency than their static bit width equivalents." @default.
- W3102403234 created "2020-11-23" @default.
- W3102403234 creator A5005095861 @default.
- W3102403234 creator A5041689212 @default.
- W3102403234 creator A5061001528 @default.
- W3102403234 creator A5068545893 @default.
- W3102403234 creator A5070019003 @default.
- W3102403234 creator A5082671220 @default.
- W3102403234 creator A5087368991 @default.
- W3102403234 date "2020-05-14" @default.
- W3102403234 modified "2023-10-16" @default.
- W3102403234 title "Bayesian Bits: Unifying Quantization and Pruning" @default.
- W3102403234 hasPublicationYear "2020" @default.
- W3102403234 type Work @default.
- W3102403234 sameAs 3102403234 @default.
- W3102403234 citedByCount "5" @default.
- W3102403234 countsByYear W31024032342021 @default.
- W3102403234 countsByYear W31024032342022 @default.
- W3102403234 crossrefType "proceedings-article" @default.
- W3102403234 hasAuthorship W3102403234A5005095861 @default.
- W3102403234 hasAuthorship W3102403234A5041689212 @default.
- W3102403234 hasAuthorship W3102403234A5061001528 @default.
- W3102403234 hasAuthorship W3102403234A5068545893 @default.
- W3102403234 hasAuthorship W3102403234A5070019003 @default.
- W3102403234 hasAuthorship W3102403234A5082671220 @default.
- W3102403234 hasAuthorship W3102403234A5087368991 @default.
- W3102403234 hasConcept C107673813 @default.
- W3102403234 hasConcept C11413529 @default.
- W3102403234 hasConcept C13280743 @default.
- W3102403234 hasConcept C154945302 @default.
- W3102403234 hasConcept C155512373 @default.
- W3102403234 hasConcept C185798385 @default.
- W3102403234 hasConcept C205649164 @default.
- W3102403234 hasConcept C28855332 @default.
- W3102403234 hasConcept C33923547 @default.
- W3102403234 hasConcept C41008148 @default.
- W3102403234 hasConceptScore W3102403234C107673813 @default.
- W3102403234 hasConceptScore W3102403234C11413529 @default.
- W3102403234 hasConceptScore W3102403234C13280743 @default.
- W3102403234 hasConceptScore W3102403234C154945302 @default.
- W3102403234 hasConceptScore W3102403234C155512373 @default.
- W3102403234 hasConceptScore W3102403234C185798385 @default.
- W3102403234 hasConceptScore W3102403234C205649164 @default.
- W3102403234 hasConceptScore W3102403234C28855332 @default.
- W3102403234 hasConceptScore W3102403234C33923547 @default.
- W3102403234 hasConceptScore W3102403234C41008148 @default.
- W3102403234 hasLocation W31024032341 @default.
- W3102403234 hasOpenAccess W3102403234 @default.
- W3102403234 hasPrimaryLocation W31024032341 @default.
- W3102403234 hasRelatedWork W2242818861 @default.
- W3102403234 hasRelatedWork W2347268398 @default.
- W3102403234 hasRelatedWork W2408177717 @default.
- W3102403234 hasRelatedWork W2779673880 @default.
- W3102403234 hasRelatedWork W2798742790 @default.
- W3102403234 hasRelatedWork W2809624076 @default.
- W3102403234 hasRelatedWork W2952111707 @default.
- W3102403234 hasRelatedWork W2952202138 @default.
- W3102403234 hasRelatedWork W2964246969 @default.
- W3102403234 hasRelatedWork W3024366773 @default.
- W3102403234 hasRelatedWork W3034933748 @default.
- W3102403234 hasRelatedWork W3053066398 @default.
- W3102403234 hasRelatedWork W3105379414 @default.
- W3102403234 hasRelatedWork W3115385953 @default.
- W3102403234 hasRelatedWork W3125935828 @default.
- W3102403234 hasRelatedWork W3186261059 @default.
- W3102403234 hasRelatedWork W3203066610 @default.
- W3102403234 hasRelatedWork W3213982839 @default.
- W3102403234 hasRelatedWork W2183038526 @default.
- W3102403234 hasRelatedWork W3087150032 @default.
- W3102403234 hasVolume "33" @default.
- W3102403234 isParatext "false" @default.
- W3102403234 isRetracted "false" @default.
- W3102403234 magId "3102403234" @default.
- W3102403234 workType "article" @default.