This talk was 🔥🔥🔥!

Here are my notes: https://t.co/2znd4zhgjb

So many takeaways that go against beliefs held by many:

✅ engineering is very important to good data science
✅ early teaming > late teaming
✅ you can do very well with few resources
✅ learn, learn, learn https://t.co/kvE2AOQAXM

— Radek Osmulski (@radekosmulski) March 24, 2021

“You Can’t Sit With Us”: Exclusionary Pedagogy in AI Ethics Education by @rajiinio @morganklauss @amironesei https://t.co/8PI11rk6wI pic.twitter.com/T3tsUAh2AM

— Rachel Thomas (@math_rachel) March 24, 2021

DeepViT: Towards Deeper Vision Transformer
pdf: https://t.co/V84QXEelAV
abs: https://t.co/PwIza18U7N pic.twitter.com/mS9WNcyZzU

— AK (@ak92501) March 23, 2021

Fast, differentiable sorting and ranking in PyTorch https://t.co/xaLlhOaejy #deeplearning #machinelearning #ml #ai #neuralnetworks #datascience #pytorch

— PyTorch Best Practices (@PyTorchPractice) March 22, 2021

"The likelihood is dead, long live the likelihood"
An article on simulation-based (aka likelihood-free) inference in the CERN newsletter by Johann Brehmer and me. https://t.co/BS8RaS8Q1E pic.twitter.com/Ga0ryfRkbB

— Kyle Cranmer (@KyleCranmer) March 22, 2021

Who needs floats? I-BERT doesn't!

I-BERT: A quantized Transformer with int-8 *only*

Get the best parameters with Transformers and use in TensorRT for a 4x (!!) speedup!

Contributed by @sehoonkim418, @amir__gholami @ZheweiYao

Try it on the hub: https://t.co/00w4evcRUe pic.twitter.com/eM1LJKgGAX

— Hugging Face (@huggingface) March 22, 2021

Being good at ML vs being good at talking online about ML are two very different things and don't often go hand in hand.

— Radek Osmulski (@radekosmulski) March 22, 2021

A year after lockdown, what have we learnt?

1) The models which suggested hundreds of thousands of people could die got the big call right, and did so early.
2) Most people can be trusted to do the right thing even it means sacrifice.
3) When governments hesitate, people die.

— Tim Harford (@TimHarford) March 22, 2021

FaceBlit: Instant Real-time Example-based Style Transfer to Facial Videos
pdf: https://t.co/Jqc6suIlqr
project page: https://t.co/NyEhVWHH65 pic.twitter.com/4CoTBZiOXI

— AK (@ak92501) March 20, 2021

Deep learning excels at unlocking the creation of impressive early demos of new applications using very little development resources.

The part where it struggles is reaching the level of consistent usefulness and reliability required by production usage.

— François Chollet (@fchollet) March 20, 2021

Avalanche: End-to-End Library for Continual Learning based on PyTorch
By @ContinualAI

Covers:
– Benchmarks
– Training
– Evaluation
– Models

Project: https://t.co/rZr2x2a7OE
GitHub: https://t.co/WzHDV632Rv pic.twitter.com/FBRmEsupDC

— ML Review (@ml_review) March 19, 2021

Today we present an approach for iterative digital game design that leverages #MachineLearning to train agents as play-testers, allowing designers to efficiently balance a game and align it with their original vision. Learn more at https://t.co/eAC9UoPBN5 pic.twitter.com/HjqRMcnnez

— Google AI (@GoogleAI) March 19, 2021