Google leaders weigh in on #data and #AI trends driving innovation https://
itworldcanada.com/article/google
-leaders-weigh-in-on-data-and-ai-trends-driving-innovation/532645
… via @itworldca
MACHINE LEARNING
-
Google Leaders Discuss Data and AI Innovation Trends
By
–
-
Microsoft MathPrompter Boosts LLM Mathematical Reasoning Performance
By
–
Microsoft’s MathPrompter Dramatically Improves LLM Performance on Mathematical Reasoning Tasks https://
syncedreview.com/2023/03/16/mic
rosofts-mathprompter-dramatically-improves-llm-performance-on-mathematical-reasoning-tasks/
… -
GPT-4 Simulating Token Prediction Through Function Calls
By
–
this works by asking GPT-4 to simulate its own abilities to predict the next token we provide GPT-4 with python functions and tell it that one of the functions acts as a language model that predicts the next token we then call the parent function and pass in the starting tokens
-
Eight Industries Leveraging Artificial Intelligence Software Today
By
–
8 Industries Benefiting from #ArtificialIntelligence Software Right Now
by Mark Allinson @mikeflache Read more: https://
buff.ly/3CVSvDI #AI #BigData #MI #DataScience #Healthcare #Finance cc: @wil_bielert @patrickgunz_ch @terenceleungsf -
Understanding Temperature in Language Models: Likelihood vs. Randomness
By
–
Temperature makes unlikely tokens more likely. Not hard math, but TLDR:
— Riley Goodside (@goodside) 16 mars 2023
temperature = 0 → always pick what the model says is most likely
temperature = 1 → sample from the distribution returned by the model
temperature = ∞ → ignore the model; pick entirely at random pic.twitter.com/dLSpyw0ha9Temperature makes unlikely tokens more likely. Not hard math, but TLDR: temperature = 0 → always pick what the model says is most likely
temperature = 1 → sample from the distribution returned by the model
temperature = ∞ → ignore the model; pick entirely at random -
Tuning GPT-4 Output: Temperature and Top-p Parameters for Creativity
By
–
At higher temperatures (>1.25) GPT-4's writing will often dissolve into nonsense, and that risk increases the longer it writes. You can also use the top-p parameter in tandem (try 0.9 to 0.99) to get more creativity without it going completely off the rails.
-
Groq Welcomes Philip Colangelo as Machine Learning Engineer
By
–
Design is making tradeoffs; innovation is figuring out a way to have it all. We're thrilled to have Philip Colangelo, Machine Learning Engineer, on our team to help us innovate. #WhyGroq #hiring #MachineLearning pic.twitter.com/mREh4M68Vs
— Groq Inc (@GroqInc) 16 mars 2023Design is making tradeoffs; innovation is figuring out a way to have it all. We're thrilled to have Philip Colangelo, Machine Learning Engineer, on our team to help us innovate. #WhyGroq #hiring #MachineLearning
-
Building the Next Generation of AI Talent
By
–
How to build the next generation of #AI talent
by @JoeWehbe7 @wef Learn more: https://
buff.ly/3bWxluA #ArtificialIntelligence #MI #Digital #DataScience #Robotics #5G #DeepLearning #DL #Industry40 cc: @ronald_vanloon @bigdata @ravikikan -
How AI Can Be Used to Fight Climate Change
By
–
How To Fight Climate Change Using #AI
by @MMinevich @forbes Learn more: https://
buff.ly/3RhBl8V #BigData #MachineLearning #ArtificialIntelligence #ML #MI #Digital #DataScience cc: @marketbuildr @amuellerml @richardsocher -
Object Recognition to Actionable Instructions: Unlocking New Possibilities
By
–
The ability to generate actionable instructions after recognizing objects in photos opens up so many possibilities