Picture for Young Jin Kim

Young Jin Kim

Division of Cardiovascular Medicine, Radcliffe Department of Medicine, University of Oxford, Department of Radiology, Severance Hospital, South Korea

Who Says Elephants Can't Run: Bringing Large Scale MoE Models into Cloud Scale Production

Nov 18, 2022
Young Jin Kim, Rawn Henry, Raffy Fahim, Hany Hassan Awadalla

Figure 1 for Who Says Elephants Can't Run: Bringing Large Scale MoE Models into Cloud Scale Production

Figure 2 for Who Says Elephants Can't Run: Bringing Large Scale MoE Models into Cloud Scale Production

Figure 3 for Who Says Elephants Can't Run: Bringing Large Scale MoE Models into Cloud Scale Production

Figure 4 for Who Says Elephants Can't Run: Bringing Large Scale MoE Models into Cloud Scale Production

* Accepted to SustaiNLP 2022 (EMNLP 2022)

Access Paper or Ask Questions

Fast Vocabulary Projection Method via Clustering for Multilingual Machine Translation on GPU

Aug 14, 2022
Hossam Amer, Young Jin Kim, Mohamed Afify, Hitokazu Matsushita, Hany Hassan Awadallah

Figure 1 for Fast Vocabulary Projection Method via Clustering for Multilingual Machine Translation on GPU

Figure 2 for Fast Vocabulary Projection Method via Clustering for Multilingual Machine Translation on GPU

Figure 3 for Fast Vocabulary Projection Method via Clustering for Multilingual Machine Translation on GPU

Figure 4 for Fast Vocabulary Projection Method via Clustering for Multilingual Machine Translation on GPU

* 12 pages, accepted at AMTA-2022 (Association for Machine Translation in the Americas Conference)

Access Paper or Ask Questions

Gating Dropout: Communication-efficient Regularization for Sparsely Activated Transformers

May 28, 2022
Rui Liu, Young Jin Kim, Alexandre Muzio, Barzan Mozafari, Hany Hassan Awadalla

Figure 1 for Gating Dropout: Communication-efficient Regularization for Sparsely Activated Transformers

Figure 2 for Gating Dropout: Communication-efficient Regularization for Sparsely Activated Transformers

Figure 3 for Gating Dropout: Communication-efficient Regularization for Sparsely Activated Transformers

Figure 4 for Gating Dropout: Communication-efficient Regularization for Sparsely Activated Transformers

* Accepted to ICML 2022

Access Paper or Ask Questions

FastFormers: Highly Efficient Transformer Models for Natural Language Understanding

Oct 26, 2020
Young Jin Kim, Hany Hassan Awadalla

Figure 1 for FastFormers: Highly Efficient Transformer Models for Natural Language Understanding

Figure 2 for FastFormers: Highly Efficient Transformer Models for Natural Language Understanding

Figure 3 for FastFormers: Highly Efficient Transformer Models for Natural Language Understanding

Figure 4 for FastFormers: Highly Efficient Transformer Models for Natural Language Understanding

* Accepted to SustaiNLP 2020 at EMNLP 2020

Access Paper or Ask Questions

Automated Quality Control in Image Segmentation: Application to the UK Biobank Cardiac MR Imaging Study

Jan 27, 2019
Robert Robinson, Vanya V. Valindria, Wenjia Bai, Ozan Oktay, Bernhard Kainz, Hideaki Suzuki, Mihir M. Sanghvi, Nay Aung, Jos$é$ Miguel Paiva, Filip Zemrak, Kenneth Fung, Elena Lukaschuk, Aaron M. Lee, Valentina Carapella, Young Jin Kim, Stefan K. Piechnik, Stefan Neubauer, Steffen E. Petersen, Chris Page, Paul M. Matthews, Daniel Rueckert, Ben Glocker

Figure 1 for Automated Quality Control in Image Segmentation: Application to the UK Biobank Cardiac MR Imaging Study

Figure 2 for Automated Quality Control in Image Segmentation: Application to the UK Biobank Cardiac MR Imaging Study

Figure 3 for Automated Quality Control in Image Segmentation: Application to the UK Biobank Cardiac MR Imaging Study

Figure 4 for Automated Quality Control in Image Segmentation: Application to the UK Biobank Cardiac MR Imaging Study

* 14 pages, 7 figures, Journal of Cardiovascular Magnetic Resonance

Access Paper or Ask Questions