Why does Stochastic Gradient Descent generalise well in deep networks?
— DeepMind (@DeepMind) February 3, 2021
Our team shows that if the learning rate is small but finite, the mean iterate of random shuffling SGD stays close to the path of gradient flow, but on a modified loss landscape https://t.co/JUAzPujWfP pic.twitter.com/NtlbVaMfLC
Scaling Laws for Transferhttps://t.co/m7Vh5aOTjx pic.twitter.com/5lw9GyOAp5
— hardmaru (@hardmaru) February 3, 2021
Turns out we can reverse engineer chunks of neural network and then write out weights by hand that reimplement it. Not sure what higher standard there is for showing we understand something. https://t.co/JKcpEolUJu
— Chris Olah (@ch402) February 2, 2021
Large, natural datasets are invaluable for training accurate, deployable systems, but are they required for driving modeling innovation? Can we use small, synthetic benchmarks instead? Our new paper asks this: https://t.co/b9WYYonQxi
— Nelson Liu (@nelsonfliu) February 2, 2021
w/ Tony Lee, @robinomial, @percyliang
(1/8) pic.twitter.com/sdhRk5UT5L
Speech Recognition by Simply Fine-tuning BERT
— AK (@ak92501) February 2, 2021
pdf: https://t.co/2kit83mnj9
abs: https://t.co/qyOosTp8Ey pic.twitter.com/brIQfVxWim
Curve Circuits — A new Distill article by @nickcammarata, @gabeeegoooh, @shancarter, @csvoss, @ludwigschubert, and @ch402. This is the sixth article in the circuits thread.https://t.co/MhDgvNVRW5 pic.twitter.com/e1QIa8qL7u
— Distill (@distillpub) February 2, 2021
Learn how to use the Determined AI platform to offload common infrastructure problems such as scaling training and hyperparameter tuning when developing #PyTorch models. https://t.co/tNYbSMZ3pz
— PyTorch (@PyTorch) February 1, 2021
I'm beyond excited to share a new book from the @TechPolicyLab. Telling Stories: On Culturally Responsive Artificial Intelligence is a collection of short stories from around the world on the social and cultural impacts of AI. https://t.co/lJHnF7XAfL pic.twitter.com/7QglaYDnWt
— Ryan Calo (@rcalo) February 1, 2021
If you want to find out what the new @spacy_io v3 is all about, check out this video I recorded with @honnibal 😇 We're walking you through some of the most exciting new features!
— Ines Montani 〰️ (@_inesmontani) February 1, 2021
📺 Watch it here: https://t.co/y0UYh7py3d pic.twitter.com/6HpnvmrAco
My awesome colleagues have now released #PyTorch version of StyleGAN2-ADA. (The initial release was in #TensorFlow )
— Ming-Yu Liu (@liu_mingyu) February 1, 2021
ADA uses a clever data augmentation to help address limit sample problems in #GAN training.https://t.co/75Yttri2KS
A useful distinction:
— Tim Harford (@TimHarford) February 1, 2021
Misinformation: incorrect or misleading information.
Disinformation: false information deliberately and often covertly spread in order to influence public opinion or obscure the truth.
Source: Merriam-Webster dictionary.
What do you call a machine learning model that perfectly predicts the training data, but does not work for unseen data?
— Christoph Molnar (@ChristophMolnar) February 1, 2021
A database
