The story behind cURL daniel.haxx.se/blog/2021/03/2…

We just released our dataset for exposure correction in photography. It is over 24,000 8-bit sRGB images rendered with different exposure settings. Each image has a properly exposed reference image for comparisons. Link: github.com/mahmoudnafifi/… #CVPR2021 #Computervision

Great summary of the standard narrative for explainable AI by Zachary Lipton

Andrew Ng inspired the Kaggle generation for the past 10 years. He is now paving the way to the post-Kaggle generation, where it's more about good data and AI governance than the machine learning models themselves. Exciting times ahead! analyticsindiamag.com/big-data-to-go…

New #drum loop added soundcloud.com/christophe-bou…

Model assertions let data scientists, developers and domain experts specify when errors in ML models may be occurring #ResponsibleAI databricks.com/blog/2021/07/2…

In scikit-learn 1.0, we decided to deprecate the sklearn.datasets.load_boston function because the design of this dataset casually assumes that people prefer to buy housing in racially segregated neighborhoods.

I will be in the San Francisco bay area next week. Happy to get suggestions of persons to meet in the #tech #SaaS #insurance #insurtech space!

As a reminder, tap the stars icon on upper right of screen to switch between latest people you follow and recommended tweets

I have unblocked everyone I blocked, apart from scammers. I recommend others do the same. Negative feedback is a good thing.

🚀 Nouvel épisode des Mini-Series Zelros sur l’IA générative appliquée à l’assurance ! Découvrez la sureté des LLM avec de vraies démos en direct ! 🎙️Christophe Bourguignat IA #Assurance #Zelros #LLM eu1.hubs.ly/H05TtbJ0

Why don't we say that a binary classifier error is an hallucination? #LLM #AllModelsAreWrongButSomeAreUseful

Young men are out-earning young women in most U.S. cities

Cost per ton to orbital space ~10,000% lower than Saturn V, is impossible. He probably meant divided by 100. Which means 99% lower.

Instruction: "Write a Python program that displays the GDP per capita for the USA and Canada from 1995 to the present, year by year, using 1995 as a base of 100." Code: colab.research.google.com/drive/1s49FJGT… Chart:

One project that impressed us was Mind The Clause by Christophe Bourguignat: mind-the-clause.lovable.app The project takes a painful and difficult to track activities and makes it easy. The creator of the Mind The Clause will receive 600 credits + $20 in AI credits + Lovable merch

Personal takeaway (maybe well known) to vibe code with Lovable (or others) on early stage projects: better iterate 10 times on 10 successive new apps, than 100 times on one same app.