Conor McDonald

Principal Data Scientist @PrimeVideo. tweeting/retweeting on Machine Learning 👨‍💻 | Forecasting 📉 | Causality 🤔. PhD, Economics.

[Albert Rapp @rappa753]

Paired bar charts suck at comparing values. The only reason they're used all the time is because they are easy to create.

But there are better alternatives that are just as easy.

Here's how to create 4 better alternatives with #rstats.

[Sebastian Raschka @rasbt]

Since self-attention is now everywhere, it's important to understand how it works.
And there is no better and more fun way than coding it from scratch!
My new article on "Understanding the Self-Attention Mechanism of Large Language Models From Scratch"
👉 https://t.co/zFsWH4orpa

[elvis @omarsar0]

If you are looking for a study guide on Transformers, look no further!

Includes notes, readings, and high-quality references to create a solid study plan for Transformers.

https://t.co/sIANvrdoU6

[Conor McDonald @conormacd]

Have just started experimenting with this and it’s awesome. Great launch 🚀

[Adam D'Angelo @adamdangelo]

Today, we are opening up public access to a new AI product we have been building called Poe. Poe lets people ask questions, get instant answers, and have back-and-forth conversations with several AI-powered bots. (1/n)

[Jim Fan @DrJimFan]

tiny-diffusion: https://t.co/FXS8xmZXQq
minGPT: https://t.co/qVyMhc0kjx
nanoGPT: https://t.co/tT0MEyODgU

Enjoy the beauty of the minimal!

[Vincent Arel-Bundock @VincentAB]

𝚖𝚊𝚛𝚐𝚒𝚗𝚊𝚕𝚎𝚏𝚏𝚎𝚌𝚝𝚜 0.9.0 is a BIG📦release! Simple yet powerful tools to help you interpret the results of over 70 classes of models in #RStats (LM, GLM, GAM, discrete choice, mixed-effects, bayesian, etc.) 🧵 on some cool new stuff. https://t.co/OwBLCPgovN

[Gael Varoquaux 🦋 @GaelVaroquaux]

⚠️A widespread confusion: calibration of predictors, as measured by expected calibration error, does not control completely that the predictor gives perfect probabilities P(y|X):

A predictor may be overconfident on some individuals and underconfident on others
🧵
1/10

[Conor McDonald @conormacd]

@DSaience Add tests to make sure that your X and y labels are aligned as intended if you are doing a lot of shuffling and splitting of the data!

[Conor McDonald @conormacd]

@ChelseaParlett 😅 hope the plot is worth it!

[Conor McDonald @conormacd]

@thejustinwelsh This is great 👍

[Conor McDonald @conormacd]

Instructor looks very promising 😊

It’s an instruction-finetuned text embedding model that can you can generate text embeddings tailored to any task and domains (e.g., science, finance, etc.) by simply providing the task instruction.

Available on 🤗

https://t.co/41wdWKrUFb

[Keerthana Gopalakrishnan @keerthanpg]

As a developer I spend a lot of hours trying to understand codebases, clicking through spiraling callstacks in mind-boggling fashion to add a change. What if AI can read the codebase instead and tell you exactly what to change where? @nuwandavek & I built https://t.co/b2rDXcCTBd

[Cameron R. Wolfe, Ph.D. @cwolferesearch]

I just found a super useful stable diffusion CLI that can efficiently perform image generation, masking, editing, outpainting, and more with no coding required. Here's some cool stuff you can do with it... 🧵[1/7]

[Conor McDonald @conormacd]

Raster Vision is an open source Python library and framework for building computer vision models on satellite, aerial, and other large imagery sets (including oblique drone imagery).

https://t.co/zhoqehRKCH

Built with PyTorch 🔥

[Conor McDonald @conormacd]

Use torchinfo to get Keras style model summaries for Pytorch 🔥🙌

https://t.co/vuw83lavMy

[soda_INRIA @[email protected] @soda_INRIA]

dirty_cat 0.4.0 beta version is out! 🎉

You can install it:

pip install dirty-cat==0.4.0b1

Some brand new features in this release:

1/5

[Conor McDonald @conormacd]

@rasbt Amazing, hopefully that’s one day soon! 😉

[Conor McDonald @conormacd]

The Machine Learning FAQs by @rasbt are a gold mine for new and experienced data scientists.

https://t.co/WMF2VcV0Yv

[Gael Varoquaux 🦋 @GaelVaroquaux]

Our benchmark of tree-based models vs deep learning for tabular data: final version.

TL:DR: from small compute budget, @scikit_learn's HistGradientBoosting is best. With finer tuning of hyperparams, XGBoost brings a gain (here n ranges from 3,000 to 10,000)