USA car accident death rates spiked more than 30% right as lockdowns went into effect. This is extraordinarily clear evidence of how lockdowns directly cause deaths.
— Teddy Petrou (@TedPetrou) April 12, 2021
This is almost certainly due to increased substance abuse. pic.twitter.com/C7NdxE9eA0
It turns out that vanilla optimizers such as Nesterov momentum and Adam work just as fine for large batch sizes.
— hardmaru (@hardmaru) April 12, 2021
Paper by @zacharynado, @jmgilmer, Chris Shallue, @_arohan_ and George Dahl conducted extensive ablations training vision and language models.https://t.co/KU8pkgAPfn https://t.co/u23KigmXTb
In Louisiana over 0.7% of the total population are prisoners. This points to a systemic failure rather than the state being a hotspot of criminal activity. Source: https://t.co/QGvN4kIONa pic.twitter.com/0Jsh8NIceb
— Simon Kuestenmacher (@simongerman600) April 12, 2021
Chart shows movies with the greatest difference between Rotten Tomatoes critic and audience ratings. Only movies with high number of reviews were included in order to ensure we have actually heard of the movies. Source: https://t.co/PkOYYbEmDk pic.twitter.com/O4x86B1aLN
— Simon Kuestenmacher (@simongerman600) April 11, 2021
Using PyTorch + NumPy? A bug that plagues thousands of open-source ML projects https://t.co/piVdQidmZH
— Andrej Karpathy (@karpathy) April 11, 2021
hah yes, a favorite super common super subtle bug 🐛. Bugs in deep learning silently make results slightly worse, pays to be v distrusting & defensive: https://t.co/5lBy4J77aS
RepVGG: Making VGG-style ConvNets Great Again
— Andrej Karpathy (@karpathy) April 11, 2021
paper: https://t.co/Y5WfgvqxHO
PyTorch code: https://t.co/ydk0RUf6JU
👌Spells out the benefits of very simple/uniform/fast (latency, not FLOPS) deployment architectures. A lot of complexity often due to optimization, not architecture. pic.twitter.com/8GliE4JDiq
New work on Efficient Transformers in RL using Actor-Learner Distillation:
— Russ Salakhutdinov (@rsalakhu) April 10, 2021
Compressing online larger “Learner model” towards a tractable “Actor model” in distributed RL setting with partially-observable environments.https://t.co/jnExWiPabS
with E. Parisotto #ICLR2021 pic.twitter.com/kC5SSRsWrn
10 Things You Need to Know About BERT and the Transformer Architecture That Are Reshaping the AI Landscape
— Sebastian Ruder (@seb_ruder) April 9, 2021
This super comprehensive post covers most things that are important in current NLP including BERT, transfer and avocado chairs 🥑
by @cathalhoran https://t.co/Hn458pkrk8 pic.twitter.com/srs4bIFrFw
So, so good. https://t.co/k4793AhiKn
— Vicki Boykis (@vboykis) April 9, 2021
Use @PrometheusIO and @Grafana to instrument your @rapidsai @dask_dev deployments and monitor how your workflows are performing on your systems. https://t.co/uCLRqyi2WJ
— RAPIDS AI (@RAPIDSai) April 9, 2021
“The data economy has to go because it is at odds with free, equal, stable, and liberal democracies.”
— Gary Marcus (@GaryMarcus) April 9, 2021
-@CarissaVeliz, speaking an important truth that few have had the courage to say out loud. https://t.co/ffyiakCGuH
So don’t get me wrong, we’re on the way out of this. Several countries are very nearly there.
— John Burn-Murdoch (@jburnmurdoch) April 9, 2021
Here is a new chart that we should all be watching over the coming months, as countries seek to follow Israel’s lead and reach the Covid endgame: pic.twitter.com/8aop6XmBBI
