Syft

This is the second release candidate for version 0.5.0!

- Recursive Serializable Plans with ResNet 18 example
- PyGrid Integration fixes
- New Mono Repo structure
- Renamed master branch to main
- Added torch==1.8.1 support
- Many more bug fixes and minor improvements

This is the first release candidate for version 0.5.0!

Happy April Fools Day - sorry, this release is NOT a joke. 😄

We have added support for the following:

- AST
- New API for support libraries
- static methods
- global variables
- dynamic methods
- enums
- Core
- Storeable refactored
- Basic Plans
- Model Centric Federated Learning
- KotlinSyft
- PySyft (as a worker lib)
- *SwiftSyft is still in development*
- *syft.js is still in development*
- CLI / Repl Support
- Remote Dataloader
- Torch Tensor
- CSV
- *Library Support
- Python
- OrderedDict
- Slice
- Iterator
- Tuple
- TenSEAL
- SyMPC
- PyDP
- PSI
- opacus
- pandas
- petlib
- PIL
- PyTorch Lightning (requires our branch)
- zsks
- numpy arrays
- Grid
- Grid Client
- FL Client
- PyTorch Support
- 1.6.x
- 1.7.x
- 1.8.x
- Python
- 3.6
- 3.7
- 3.8
- 3.9
- OS
- Mac
- Windows
- Linux
- Updated Duet Logo
- New Logging System
- Integration Tests
- Duet Notebook Tests
- Benchmark Tests
- Examples
- DCGAN
- MNIST
- MNIST Lightning
- SuperResolution
- VAE
- Model Centric Federated Learning
- Homomorphic Encryption
- Vertical Learning
- Private Set Intersection
- Secure Multi Party Compute
- Privacy AI Series (Course 2)

\* Please know that library support is still incomplete. This release is intended to support the basic, minimum requirements for the below libraries.

Acknowledgements

While this isn't an exhaustive list, we'd like to thank the following people for contributing to this release!

Adam J Hall
Animesh Gupta
Anshuman Singh
Anubhav Raj Singh
Ariann Farias
ArtistBanda
Arturo Marquez Flores
Avinash Swaminathan
Ayoub Benaissa
Ben Fielding
Bogdan Cebere
Chinmay Shah
Chirag Gomber
Chitresh Goel
Daniele Gadler
George-Cristian Muraru
Gokkulnath TS
H4LL
Harkirat Singh
Hideaki Takahashi (Syumei)
Ionesio Junior
Irina Bejan
JMLourier
Jasopaum
Jirka Borovec
Joel Lee
José Benardi de Souza Nunes
Koen van der Veen
Lee Yi Jie Joel
Madhava Jay
Nabanita Dash
Nahua
Nicolas Remerscheid
Param Mirani
Patrick Cason
Plamen Hristov
Praful932
Raghav Prabhakar
Rima
Robert Perrotta
Syumei
Théo Ryffel
Tudor Cebere
Vaibhav Vardhan
Vova Manannikov
Wansoo Kim
Zarreen Naowal Reza
akshay.verma
animesh-007
avinsit123
curtismitchell
dnabanita7
gleec14
jmaunon
koukyosyumei
lazy
marcalph
nahuakang
rahul
rajatrc1705
root
shubhendra
syumeikoukyo
tchaton
uid42
xutongye

PySyft 0.3.0 Release Notes

Welcome to the next evolution of PySyft. Today we're releasing PySyft 0.3.0, which is a complete rewrite over the 0.2.x product line. Since this is a complete rewrite of the codebase, we strongly suggest that you use 0.3.0 if you're just getting started with PySyft. If you're an existing user of PySyft 0.2.x, there will still be some support for that product line but we encourage you to upgrade. As always, we are happy to assist in this process over Slack.

For those interested in trying out PySyft 0.3.0, [we suggest you start with our "Duet" series of demos](https://duet.openmined.org). All future releases will include a detailed breakdown of breaking changes, fixes, bug patches, and improvements.

Patched

Patched

> TL;DR: An incremental release for the benefit of mobile workers and pygrid. A new backend for encrypted computation.

PySyft 0.2.8 Release Notes