CvT: Introducing Convolutions to Vision Transformers
— AK (@ak92501) March 30, 2021
pdf: https://t.co/59l2nRLrLe
abs: https://t.co/xU9s1p2N45 pic.twitter.com/CEoaWE7peo
CvT: Introducing Convolutions to Vision Transformers
— AK (@ak92501) March 30, 2021
pdf: https://t.co/59l2nRLrLe
abs: https://t.co/xU9s1p2N45 pic.twitter.com/CEoaWE7peo
ViViT: A Video Vision Transformer
— AK (@ak92501) March 30, 2021
pdf: https://t.co/3veTg7iQZ1
abs: https://t.co/hIQLtWXKcD pic.twitter.com/2zpt2Gl6Lc
PixelTransformer: Sample Conditioned Signal Generation
— AK (@ak92501) March 30, 2021
pdf: https://t.co/ncExK3yiid
abs: https://t.co/asdtHwlc8T
project page: https://t.co/0FNfQCy0yA pic.twitter.com/JTNNLapVWn
“The core problem is we build recruiting policies around the mean student. Yet hiring is about the tail.” https://t.co/EM14Lnlyql
— hardmaru (@hardmaru) March 29, 2021
A new fun quick short story on AI: "Forward Pass" 🧠✨ https://t.co/7ROcrHBqzI
— Andrej Karpathy (@karpathy) March 28, 2021
Out of Distribution Generalization in Machine Learning
— hardmaru (@hardmaru) March 28, 2021
(Martin Arjovsky's PhD Thesis)https://t.co/HDbDbNhPcm
New code walkthrough on https://t.co/m6mT8SrKDD: using a siamese network to learn to estimate how similar two images look like -- trained on the "Totally Look Like" dataset. Super readable and nicely explained. Created by @hazemessamm & @svpino
— François Chollet (@fchollet) March 26, 2021
Check it: https://t.co/9wruW7oyDL pic.twitter.com/X7RLmsxiEC
High-Fidelity Pluralistic Image Completion with Transformers
— AK (@ak92501) March 26, 2021
pdf: https://t.co/39h0sEFPeu
abs: https://t.co/NqtgQ6ZFHe
project page: https://t.co/y0p6t9GMnR pic.twitter.com/AO43jF8SVj
An Image is Worth 16x16 Words, What is a Video Worth?
— AK (@ak92501) March 26, 2021
pdf: https://t.co/vf53DlyFwy
abs: https://t.co/2wG2qCT53o
github: https://t.co/ZGpmrFK7qy pic.twitter.com/2yUWK0tP5g
Introducing torch.profiler! New PyTorch Profiler collects both GPU and framework related info, correlates them, performs automatic detection of bottlenecks in the model, generates recommendations on how to resolve these bottlenecks, and visualize.
— PyTorch (@PyTorch) March 25, 2021
Read 👉https://t.co/Ottly5CtF4 pic.twitter.com/zkFCts1lzn
GPT-3 is currently generating an average of **4.5 billion words per day**: https://t.co/pqwhce8rD3
— Greg Brockman (@gdb) March 25, 2021
I am often asked why I insist on complicating a plot by overlaying a violin plot on top of a box plot when the latter already gives a good visual summary of the distribution.
— Indrajeet Patil (@patilindrajeets) March 25, 2021
This gif provides a reason.
For a more nuanced argument, see:https://t.co/GhGmqxRoLi#rstats #dataviz pic.twitter.com/Qk1lQMuBJm