You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
Tweetorial: high-performance multi-GPU training with Keras. The only thing you need to do to turn single-device code into multi-device code is to place your model construction function under a "distribution strategy" scope, like this.pic.twitter.com/hfBSGwcRSm
There are multiple distribution strategies available. Most of the time, you will use MirroredStrategy (replicates your model on each available GPU, send a sub-batch to each replica at every training step, and keeps the replicas in sync after processing every step) or TPUStrategy.
I tried multipleStrategy on multiple machines but didn’t work out. The ip adress is recognized by both but then the terminal hangs... a short tuto on requirements for this functionality would be great 
Except memory issues it always worked quite well for me. What does the ip address have to do with it?
Is there any way to get reproducible results with GPU, especially mirrored strategy? Conventional wisdom (and personal experience) seems to be that it's only possible with pure-CPU learning.
This isn't possible at this time.
wow nice 
functionalities being added to @TensorFlow & Keras. I wish I had another GPU to try this.pic.twitter.com/dLlxPdbtSV
Always works like a charm
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
