Tag: w_code

#PyTorch implementation and pretrained models for

RepVGG: Making VGG-style ConvNets Great Againhttps://t.co/F4TnXuXr1e

- multi-branch topology at training
- simple 3x3 convs & ReLU at inference
- >80% top-1 accuracy on ImageNet
- 83% faster than ResNet-50 on NVIDIA 1080Ti pic.twitter.com/hwN60o0T78

— Alexandr Kalinin (@alxndrkalinin) January 19, 2021

Indeed, the 2nd half is highly worthwhile - inspired a lot of our current work on GTN. Do yourself a favor: https://t.co/Q6YARJyuFV

The GTN framework and corresponding preprint are here:

Code: https://t.co/5QkN5v3RFD
Paper: https://t.co/ajaWaKDr2B https://t.co/CCorfDB7Ag

— Awni Hannun (@awnihannun) January 17, 2021

Introducing ToTTo, an open domain table-to-text generation dataset, a challenging benchmark for high-precision text generation, and a controlled text generation task that can be used to assess model “hallucination”. Read more and grab the code at https://t.co/9NaFcZvFo5

— Google AI (@GoogleAI) January 15, 2021

🚀Excited to release Robustness Gym, a new Python evaluation toolkit for evaluating the robustness of NLP models, as part of a collaboration between Stanford, Salesforce Research and UNC Chapel-Hill.

Paper: https://t.co/WjcfdE42x1
Code: https://t.co/1V4fAtJvcD

pip install!

— Karan Goel (@krandiash) January 14, 2021

Re-labeling ImageNet: from Single to Multi-Labels, from Global to Localized Labels
pdf: https://t.co/eb1qvqaO7I
abs: https://t.co/rw45sireIH
github: https://t.co/C6Po5bcKKt pic.twitter.com/KFKVBEbsfA

— AK (@ak92501) January 14, 2021

Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity
pdf: https://t.co/0i6fcOuy4X
abs: https://t.co/AUKgennqZy
github: https://t.co/8QD4sJ2ckE pic.twitter.com/iDPDXj4bRR

— AK (@ak92501) January 12, 2021

I use #gganimate in #Rstats to animate cell cycle genes from our 10k gene MERFISH PNAS paper. It's like a #singlecell loading icon!

Code: https://t.co/Lx3x38ipQQ
Paper: https://t.co/1wQIQuGEZR

🥂#BioInformatics #DataScience #DataVisualization #dataviz #bioviz #funwithcode pic.twitter.com/SD15yoIyFe

— Dr. Jean Fan (@JEFworks) December 28, 2020

Trained stylegan on high resolution Islamic mosaic art. You can find all resources below and even do interpolation and other fun stuff.

GitHub: https://t.co/j2jWgW8Mdv
Colab: https://t.co/AziONqbdI2 pic.twitter.com/ekPVIELSr0

— Zaid زيد (@zaidalyafeai) December 27, 2020

Efficient transformer architectures for vision.
Open source. https://t.co/MTCgPRgr6k

— Yann LeCun (@ylecun) December 24, 2020

Training data-efficient image transformers & distillation through attention
pdf: https://t.co/52o44tGC5G
abs: https://t.co/XohbkS7kq5
github: https://t.co/KpkMSwfKdp pic.twitter.com/KP24f00vT4

— AK (@ak92501) December 24, 2020

In July, we shared a state of the art voice model that separates up to five different voices simultaneously speaking, improve the ability to reduce background noise, and enhance communication across a variety of applications. Code now on @PyTorch: https://t.co/4Trw7a8i5T pic.twitter.com/3CFeGNHIpB

— Facebook AI (@facebookai) December 22, 2020

Hypersim: Photorealistic Synthetic Dataset for Indoor Scene Understanding

Photorealistic synthetic scenes have the advantage of giving us as many ground truth layers as we want to train an ML system. But is it enough for sim2real?https://t.co/U4qXXKn6qFhttps://t.co/rTnWT4XvEE pic.twitter.com/aRa5K6unWh

— hardmaru (@hardmaru) December 22, 2020