Introducing a novel mixture-of-experts routing algorithm, called Expert Choice, that can achieve optimal load balancing between experts while allowing heterogeneous token-to-expert mapping. Learn how it’s done at https://
goo.gle/3OdpO9t
MACHINE LEARNING
-
Expert Choice: Novel Mixture-of-Experts Routing Algorithm
By
–
-
Groq Compiler: Fast ML/AI Model Compilation Without Kernel Libraries
By
–
We run 100's of #ML/#AI models on Groq #Compiler, with no need for kernel libraries. We can tune your workload, with #software tools like GroqFlow and Groq #API. See for yourself–come by booth 3047 at @Supercomputing
, bring your model, and let's talk time-to-compile! #SC22 #HPC -
AI Cloud Platform Converts Live Sports Action into Social Media Highlights
By
–
This week on The #AI Podcast, Amos Berkovich covers how the @WSC_Sports AI cloud platform uses #deeplearning to turn in-game action into social media-ready highlights in real time. https://
nvda.ws/3Gjf5Z9 -
Brain-inspired computational models enhancing machine learning algorithms
By
–
Brain-inspired = how computational models of brains can inspire "better" machine learning algorithms. See here for more context: https://
youtu.be/IlliqYiRhMU -
Ramin Hasani on Liquid Neural Nets and Sequence Modeling
By
–
Hi there! It’s Ramin Hasani @raminmh
! Over the next 24H, I’m taking over @MIT_CSAIL
’s Twitter! I’m a research affiliate at CSAIL. I design brain-inspired robust deep learning algorithms! Ask me anything about sequence modeling, time series, robots, & liquid neural nets, here -
TensorFlow and Keras Deep Learning Tutorial Without PhD Required
By
–
TensorFlow, Keras, and deep learning, without a Ph.D. What you'll learn: • What is a neural network
• How to build one
• Learning rate schedules
• How to build convolutional neural networks
• How to use regularization
• What is overfitting https://
codelabs.developers.google.com/codelabs/cloud
-tensorflow-mnist/#0
… -
Eliminating Gender Bias in Sentiment Analysis Systems
By
–
De-biasing sexist sentiment classifiers pic.twitter.com/QBLJtIknN6
— IBM Data, AI & Automation (@IBMData) 16 novembre 2022De-biasing sexist sentiment classifiers
-
Testing Stock Prediction Models for Security Flaws
By
–
Testing stock prediction models for security flaws 📈 pic.twitter.com/XO9AQCglr1
— IBM Data, AI & Automation (@IBMData) 16 novembre 2022Testing stock prediction models for security flaws
-
Using AI-Generated Images to Enhance Machine Translation Models
By
–
Training models on “hallucinated images”, like @OpenAI’s Dall-E, to improve machine translation pic.twitter.com/TM9hvBvixr
— IBM Data, AI & Automation (@IBMData) 16 novembre 2022Training models on “hallucinated images”, like @OpenAI
’s Dall-E, to improve machine translation -
Probabilistic Generative Modeling Enhances Machine Design Efficiency
By
–
Designing machines using probabilistic generative modeling to bring more efficiency into the design process pic.twitter.com/h8BHvqfOHB
— IBM Data, AI & Automation (@IBMData) 16 novembre 2022Designing machines using probabilistic generative modeling to bring more efficiency into the design process