Yay! differentiable FFT! https://t.co/BXGOFhFih7
— Yann LeCun (@ylecun) March 10, 2021
Yay! differentiable FFT! https://t.co/BXGOFhFih7
— Yann LeCun (@ylecun) March 10, 2021
Introducing VISSL (https://t.co/iBEpmCi09R) - a library for reproducible, SOTA self-supervised learning for computer vision! Over 10 methods implemented, 60 pre-trained models, 15 benchmarks, and counting. pic.twitter.com/ZZMd8DpHBD
— PyTorch (@PyTorch) March 9, 2021
FairScale, a PyTorch extension for efficient large scale training, is releasing FullyShardedDataParallel, which shards model params across GPUs (+offload to CPU). Details: https://t.co/xshPfLeXyr. Inspired by DeepSpeed/@MSFTResearch, and made by @myleott @m1nxu @sam_shleifer pic.twitter.com/1ICMsJwtUP
— PyTorch (@PyTorch) February 25, 2021
Ray is an open-source library for parallel and distributed Python. It can be paired with #PyTorch to rapidly scale machine learning applications. Learn more below: https://t.co/UhfPxSEoWm
— PyTorch (@PyTorch) February 19, 2021
This is nice! I didn't know about this autograd refresh via the new functional API. So, among other things, you can now compute partial derivatives / gradients from Python function as opposed to Python variables. Beyond the performance improvements, good for code organization. https://t.co/EgUv03ZeiZ pic.twitter.com/cIRk0D7puk
— Sebastian Raschka (@rasbt) February 8, 2021
Lightning made the mistake of relying on inheritance instead of callbacks, not realizing that inheritance has fundamental well-known limitations for extensibility.
— Jeremy Howard (@jeremyphoward) February 5, 2021
They even boasted about this.
Later, they were forced to add callbacks. Oops. pic.twitter.com/w7g7XnRl6t
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
#PyTorch implementation and pretrained models for
— Alexandr Kalinin (@alxndrkalinin) January 19, 2021
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
Here's a boatload of model implementations for Deep Learning on Graphs https://t.co/KZVJ6wz76A @GraphDeep https://t.co/Ne2v3rNguI
— Alex Smola (@smolix) January 8, 2021
FairTorch secured the first place at the Summer Hackathon 2020 in the Responsible AI category. The tool allows you to add a fairness constraint to your model with a few lines of code, using the fairness criteria provided in the library. https://t.co/40svXCVOut
— PyTorch (@PyTorch) January 6, 2021
[Release] TorchServe 0.3 includes new integrations for: Kubelow/KFServing, Captum for model interpretability, MLflow-torchserve plugin, Google Kubernetes and Azure Kubernetes Engine support, Native Windows support, gPRC interfaces for all APIs, and more. https://t.co/Do52ONJkFB
— PyTorch (@PyTorch) January 5, 2021
HyperLSTM PyTorch implementation https://t.co/c59I64qqGP
— /MachineLearning (@slashML) January 3, 2021