16 hours ago#MLefficieny — Meta MultiRay to manage Deep Learning at scaleMultiRay is a platform developed by Meta (Facebook) to provide Deep Learning applications for a variety of Use Cases while at the same time keeping the costs at a reasonable level. MulitRay is used in more than 125 applications at Meta. Here we are focusing on language tasks. However, it…Machine Learning6 min readMachine Learning6 min read
19 hours ago#MLefficiency — Optimizing transformer models for efficiencySince models are growing in the number of parameters, it increases the computational requirements at the same time. However, during inference, you want to allocate as less compute resources as possible for a service to run 24/7, simply to keep the operational costs low. Therefore, an active field of research…Machine Learning5 min readMachine Learning5 min read
Published in kimesta AI·Nov 14The power of embeddingsCombining large language models with image understanding — Three Takeaways Definiton of the term embedding CLIP as a powerful example for learning embeddings Embeddings are powerful in a variety of Deep Learning areas. Embeddings Embeddings are representations of information. The representation captures the semantic meaning of what is being embedded, making it robust for many industry applications.Clip4 min readClip4 min read
Apr 11, 2019Member-onlyWeight Imprinting on the Coral Edge TPU [Part 1]This article was co-written with my fellow mate and colleague Johannes. Part 1 describes the theory of Weight Imprinting, Part 2 continues with the implementation of Weight Imprinting on the Coral Edge TPU. Motivation Deep Learning is an amazingly powerful tool that is developing at an insane speed. It is also…Machine Learning8 min readMachine Learning8 min read
![Weight Imprinting on the Coral Edge TPU [Part 1]](https://miro.medium.com/fit/c/224/224/1*-EKKnURqMZgBbxm0-jj5EA.jpeg)
![Weight Imprinting on the Coral Edge TPU [Part 1]](https://miro.medium.com/fit/c/160/112/1*-EKKnURqMZgBbxm0-jj5EA.jpeg)
