Matches in SemOpenAlex for { <https://semopenalex.org/work/W3035604854> ?p ?o ?g. }
- W3035604854 abstract "Large-batch training is an efficient approach for current distributed deep learning systems. It has enabled researchers to reduce the ImageNet/ResNet-50 training from 29 hours to around 1 minute. In this paper, we focus on studying the limit of the batch size. We think it may provide a guidance to AI supercomputer and algorithm designers. We provide detailed numerical optimization instructions for step-by-step comparison. Moreover, it is important to understand the generalization and optimization performance of huge batch training. Hoffer et al. introduced ultra-slow diffusion theory to large-batch training. However, our experiments show contradictory results with the conclusion of Hoffer et al. We provide comprehensive experimental results and detailed analysis to study the limitations of batch size scaling and ultra-slow diffusion theory. For the first time we scale the batch size on ImageNet to at least a magnitude larger than all previous work, and provide detailed studies on the performance of many state-of-the-art optimization schemes under this setting. We propose an optimization recipe that is able to improve the top-1 test accuracy by 18% compared to the baseline." @default.
- W3035604854 created "2020-06-19" @default.
- W3035604854 creator A5008491549 @default.
- W3035604854 creator A5010841999 @default.
- W3035604854 creator A5024120345 @default.
- W3035604854 creator A5029277113 @default.
- W3035604854 creator A5065361552 @default.
- W3035604854 creator A5076825233 @default.
- W3035604854 date "2020-06-15" @default.
- W3035604854 modified "2023-09-24" @default.
- W3035604854 title "The Limit of the Batch Size." @default.
- W3035604854 cites W1498436455 @default.
- W3035604854 cites W1522301498 @default.
- W3035604854 cites W1598866093 @default.
- W3035604854 cites W2039417226 @default.
- W3035604854 cites W2146502635 @default.
- W3035604854 cites W2155894447 @default.
- W3035604854 cites W2194775991 @default.
- W3035604854 cites W2518108298 @default.
- W3035604854 cites W2523060838 @default.
- W3035604854 cites W2617242334 @default.
- W3035604854 cites W2618003483 @default.
- W3035604854 cites W2622263826 @default.
- W3035604854 cites W2755682530 @default.
- W3035604854 cites W2757910899 @default.
- W3035604854 cites W2766164908 @default.
- W3035604854 cites W2769856846 @default.
- W3035604854 cites W2884711234 @default.
- W3035604854 cites W2888206291 @default.
- W3035604854 cites W2900167092 @default.
- W3035604854 cites W2901541570 @default.
- W3035604854 cites W2902280036 @default.
- W3035604854 cites W2920668770 @default.
- W3035604854 cites W2921416272 @default.
- W3035604854 cites W2926655273 @default.
- W3035604854 cites W2945697643 @default.
- W3035604854 cites W2949935872 @default.
- W3035604854 cites W2950300355 @default.
- W3035604854 cites W2964054038 @default.
- W3035604854 cites W2967791890 @default.
- W3035604854 cites W2973727699 @default.
- W3035604854 cites W2974008169 @default.
- W3035604854 cites W2994689640 @default.
- W3035604854 cites W2998944890 @default.
- W3035604854 cites W3025935268 @default.
- W3035604854 hasPublicationYear "2020" @default.
- W3035604854 type Work @default.
- W3035604854 sameAs 3035604854 @default.
- W3035604854 citedByCount "2" @default.
- W3035604854 countsByYear W30356048542021 @default.
- W3035604854 crossrefType "posted-content" @default.
- W3035604854 hasAuthorship W3035604854A5008491549 @default.
- W3035604854 hasAuthorship W3035604854A5010841999 @default.
- W3035604854 hasAuthorship W3035604854A5024120345 @default.
- W3035604854 hasAuthorship W3035604854A5029277113 @default.
- W3035604854 hasAuthorship W3035604854A5065361552 @default.
- W3035604854 hasAuthorship W3035604854A5076825233 @default.
- W3035604854 hasConcept C119857082 @default.
- W3035604854 hasConcept C120665830 @default.
- W3035604854 hasConcept C121332964 @default.
- W3035604854 hasConcept C127413603 @default.
- W3035604854 hasConcept C134306372 @default.
- W3035604854 hasConcept C146978453 @default.
- W3035604854 hasConcept C151201525 @default.
- W3035604854 hasConcept C154945302 @default.
- W3035604854 hasConcept C172658912 @default.
- W3035604854 hasConcept C177148314 @default.
- W3035604854 hasConcept C192209626 @default.
- W3035604854 hasConcept C199360897 @default.
- W3035604854 hasConcept C204323151 @default.
- W3035604854 hasConcept C2524010 @default.
- W3035604854 hasConcept C33923547 @default.
- W3035604854 hasConcept C41008148 @default.
- W3035604854 hasConcept C69357855 @default.
- W3035604854 hasConcept C97355855 @default.
- W3035604854 hasConcept C99844830 @default.
- W3035604854 hasConceptScore W3035604854C119857082 @default.
- W3035604854 hasConceptScore W3035604854C120665830 @default.
- W3035604854 hasConceptScore W3035604854C121332964 @default.
- W3035604854 hasConceptScore W3035604854C127413603 @default.
- W3035604854 hasConceptScore W3035604854C134306372 @default.
- W3035604854 hasConceptScore W3035604854C146978453 @default.
- W3035604854 hasConceptScore W3035604854C151201525 @default.
- W3035604854 hasConceptScore W3035604854C154945302 @default.
- W3035604854 hasConceptScore W3035604854C172658912 @default.
- W3035604854 hasConceptScore W3035604854C177148314 @default.
- W3035604854 hasConceptScore W3035604854C192209626 @default.
- W3035604854 hasConceptScore W3035604854C199360897 @default.
- W3035604854 hasConceptScore W3035604854C204323151 @default.
- W3035604854 hasConceptScore W3035604854C2524010 @default.
- W3035604854 hasConceptScore W3035604854C33923547 @default.
- W3035604854 hasConceptScore W3035604854C41008148 @default.
- W3035604854 hasConceptScore W3035604854C69357855 @default.
- W3035604854 hasConceptScore W3035604854C97355855 @default.
- W3035604854 hasConceptScore W3035604854C99844830 @default.
- W3035604854 hasLocation W30356048541 @default.
- W3035604854 hasOpenAccess W3035604854 @default.
- W3035604854 hasPrimaryLocation W30356048541 @default.
- W3035604854 hasRelatedWork W1599080376 @default.
- W3035604854 hasRelatedWork W2340405466 @default.