The focus for this MeetUp's main talks will be on understanding how to make your Deep Learning faster, which is something practitioners are probably eager to learn. Even if you're not actively doing Deep Learning yet, understanding the pain-points is pretty important too.
Planned Talks :
"Optimising your TensorFlow installation" - Jason Zaman
Jason is a Gentoo Linux developer and maintainer on the SELinux Project. He maintains Tensorflow and Android Studio on Gentoo and as part of the Hardened project has extensive experience with toolchains. He is also an active member of the Open Source community and speaks regularly on Android and SELinux.
He will be speaking about optimizing Tensorflow and using it to the
fullest. He will also be covering the Bazel build system that Tensorflow uses and some CPU and compiler extensions that when taken advantage of can give significant speedups.
"Faster Machine Learning in the Cloud" - Sam Witteveen
Google has recently come out with better tooling (eg: Keras 1.9), and optimised machine images for use in their cloud. Sam will talk about how you can use these improvements to get the most out of your models.
"2x+ speed-ups using 16-bit GPU ops" - Martin Andrews
The newer Nvidia GPUs include logic for specialised 16-bit operations. For the P100s, that means (theoretically) 2x throughput, and 50% memory usage. And the V100s can get 10x+ performance if you can make use of their specialised vector units. However, there are some tricks and traps in getting this to work! Martin will explain the basic issues, and how to make sure your models can make use of the latest hardware.
If you have something that you'd like to present in a welcoming environment, please let us know by suggesting yourself below... We're enthusiastic about Lightning Talks, which are a great way of showing people cool stuff that you've been working on, without the (imagined) pressure of the "Full Presentation".
Talks will start at 7pm (A/V equipment permitting) and end at around 9pm, at which point people normally come up to the front for a bit of a chat with each other, and the speakers.
As always, we're actively looking for more speakers for future events - both '30 minutes long-form', and lightning talks. For the lightning talks, we welcome folks to come and talk about something cool they've done with TensorFlow and/or Deep Learning for 5-10mins (so, if you have slides, then #max=10). We believe that the key ingredient for the success of a Lightning Talk is simply the cool/interesting factor. It doesn't matter whether you're an expert or and enthusiastic beginner: Given the responses we have had, we're sure there are lots of people who would be interested to hear what you've been playing with. Please suggest yourself here :