cntk

CNTK - The Microsoft Cognitive Toolkit

[cntk @mscntk]

In partnership with edX we have released a new course "Deep Learning Explained": https://t.co/PkASsXi6CL

[Emad Barsoum @EmadBarsoumPi]

New high performance inference engine based on @onnxai

https://t.co/h4XKQaPiej

[cntk @mscntk]

We just released CNTK 2.6: ONNX 1.2.2 support, .Net Core 2.0, better performance, convolution on sequence axis. bug fixes and more.

https://t.co/1HoAY3uXLx

[cntk @mscntk]

CNTK 2.5 is out: more granular profiling with chrome trace format, GRU and LSTM support for ONNX, perf improvement, multiple learners for distributed training and many more. Check out the details:

https://t.co/AYC8da0pPx

[cntk @mscntk]

https://t.co/VlYndc3Hp4

[cntk @mscntk]

CNTK 2.4 is out, Volta fp16 support, more ONNX OPs, perf improvement...etc. Details in: https://t.co/AYC8da0pPx

[cntk @mscntk]

CNTK 2.3.1 release with ONNX support: https://t.co/AYC8da0pPx

[cntk @mscntk]

https://t.co/QaVS3gzO8Z

[cntk @mscntk]

New super-resolutions tutorials!
Training:
https://t.co/g0hCE5pRYk
and evaluation:
https://t.co/ieUG9XsZLN

[cntk @mscntk]

CNTK 2.3 has been released. Highlights:
Better ONNX support.
NCCL2 for distributed training.
Improved C# API.
Network Optimization API.
Faster Adadelta for sparse gradients.
Various performance improvements, and more
https://t.co/kFjY3QGP58

[cntk @mscntk]

CNTK September interation plan has been posted: https://t.co/fRyeRZae9X

Also, CNTK R-Binding is now public: https://t.co/x1ejKj8t5u

[cntk @mscntk]

CNTK 2.2 released! https://t.co/Nf8jh4sdTl

C# and R bindings, multigpu Keras and many more new features and performance improvements.

[cntk @mscntk]

CNTK, Caffe2 and PyTorch model interoperability!
https://t.co/p92o1fMhan

[cntk @mscntk]

CNTK model eval in UWP!
https://t.co/xJKA3S1qMk

[cntk @mscntk]

CNTK powers the latest breakthrough in speech recognition https://t.co/pVemxRfVSg

[Microsoft Research @MSFTResearch]

Microsoft researchers reach new industry milestone, substantially increasing #speechrecognition accuracy https://t.co/t75Qbtrnyx

[cntk @mscntk]

CNTK 2.1 released!
https://t.co/GwXsTnxfCF

[cntk @mscntk]

Slides for CNTK tutorial at CVPR 2017 has been posted: https://t.co/ZLrGRjSS1L

The tutorial is scheduled at 1:30-5:00pm on July 26.

[cntk @mscntk]

ICYMI: Keras Backend Benchmark (updated to include the NVidia M60). CNTK backend 2x - 4x faster on RNNs. https://t.co/BkBrG9Scal

[Kaggler TV @KagglerTV]

Keras Backend Benchmark: Theano vs. TensorFlow vs. CNTK https://t.co/boQu85Jmu9

[cntk @mscntk]

CNTK July iteration plan has been released on GitHub:
https://t.co/ykvy0JPseg

[cntk @mscntk]

How to model complex data with CNTK:
https://t.co/0GiCDFK7vE

[cntk @mscntk]

The (still beta) CNTK Keras backend, already speeds up your RNNs by a factor of 2 to 4 https://t.co/tSIniVDbSZ