✨🗣️ Have you heard a lot about BERT, but are still trying to grok its architecture and how it represents linguistic information internally?
— 👩💻 DynamicWebPaige (@DynamicWebPaige) February 18, 2020
👉 Check out this explanatory paper and blog post from @viegasf and team:
📄 https://t.co/iYCeE4e9si
👩💻 https://t.co/bwMd2i4UVy pic.twitter.com/OwgVHuxdzt
"This paper introduces a new graph-based recurrent retrieval approach, which retrieves reasoning paths over the @Wikipedia graph to answer multi-hop open-domain questions."https://t.co/MnUbqKIetH
— 👩💻 DynamicWebPaige (@DynamicWebPaige) February 17, 2020
Experimental results show state-of-the-art results in 3 open-domain QA data sets! pic.twitter.com/2BQL3TjPck
Transformers as Soft Reasoners over Language
— Thomas (@evolvingstuff) February 17, 2020
"we explore whether transformers can similarly learn to reason (or emulate reasoning), but using rules expressed in language, thus bypassing a formal representation."https://t.co/5uib4bRPud
Datasets and demo: https://t.co/wKxgdCpHiE pic.twitter.com/lffXdgIKTe
self-distillation takes many useful forms in ML research (Q-learning, quantization, teacher-student architectures). Awesome fundamental work! https://t.co/6trYyiANxZ
— Eric Jang 🇺🇸🇹🇼 (@ericjang11) February 16, 2020
GANILLA: Generative Adversarial Networks for Image to Illustration Translationhttps://t.co/iHjYU1mzVx
— hardmaru (@hardmaru) February 16, 2020
code/pre-trained models/dataset https://t.co/FSRNdRnyc0
Style transfer using Ghibli anime: https://t.co/KyL6ubSUmT pic.twitter.com/fk2uWIO1k8
Rapid unsupervised learning progress thanks to contrastive losses, approaching supervised learning!
— Oriol Vinyals (@OriolVinyalsML) February 14, 2020
-40% Multitask SSL https://t.co/3HXS53l71v (2017)
-50% CPC https://t.co/k55jv1bAj3 (2018)
-70% AMDIM/MOCO/CPCv2/etc (2019)
-76.5% SimCLR https://t.co/y7WitlTOl7 (2020, so far) pic.twitter.com/z1Q1yPi9pO
1/5 Self-Distillation loop (feeding predictions as new target values & retraining) improves test accuracy. But why? We show it induces a regularization that progressively limits # of basis functions used to represent the solution. https://t.co/570qXFmlGj w/@farajtabar P.Bartlett pic.twitter.com/b79Q6ZSxlS
— Hossein Mobahi (@TheGradient) February 14, 2020
This might be the first academic paper of @dask_dev with @RAPIDSai acceleration, along with many other #Python machine learning libraries. Thank you @cjnolet and @rasbt for all your hard work, and of course, the RAPIDS powered community for giving us things to write about. https://t.co/ENYzG0cB0C
— Joshua Patterson (@datametrician) February 13, 2020
Improved Consistency Regularization for GANs
— roadrunner01 (@ak92501) February 13, 2020
pdf: https://t.co/IJki7WDD9T
abs: https://t.co/h1BPZ0KKWd pic.twitter.com/mDLyq1NbYS
Super excited to share a preprint of a relatively comprehensive review "Machine Learning in Python: Main developments and technology trends in data science, machine learning, and artificial intelligence" @datametrician, @cjnolet and I have been working on! https://t.co/7GbRy3TyTC pic.twitter.com/G1aWm4QrgE
— Sebastian Raschka (@rasbt) February 13, 2020
SketchTransfer: A Challenging New Task for Exploring Detail-Invariance and the Abstractions Learned by Deep Networks
— hardmaru (@hardmaru) February 12, 2020
If we train a neural net to classify CIFAR10 photos but also give it unlabelled QuickDraw doodles, how well can it classify these doodles? https://t.co/dh4Naces5t pic.twitter.com/YwjFcmTR17
This is great work that collects corpora and evaluates models for two extremely low-resource languages spoken in Africa 🌍, Twi and Yoruba.
— Sebastian Ruder (@seb_ruder) February 11, 2020
Link to the paper: https://t.co/Cm5YhzTbBL https://t.co/mjAvuVyPjE
