Joeri Hermans

It's actually a long time since I've tried it (a lot has changed during that time). However, if I recall correct, I'm using dense numpy vectors / matrices at a...

How many partitions does your DataFrame or RDD consist of?

Is your Keras version up-to-date?

Hi, Thanks for your contribution! Before merging, could you provide a usage example of this particular feature? Since it might impact performance to some extend. Joeri

If the machines have GPU's, it should automatically allocate the GPU's during training. Let me know if it doesn't work. Personally I wouldn't use DOWNPOUR due to "implicit momentum" as...

If the community is interested, I can make a Spark-less script which does the thing you want. Should be rather trivial to implement.

I wouldn't use synchronous SGD since it implicitly increases the size of a mini-batch :) If you want we can work together (I don't have multiple GPU's to test atm),...

Hi Christaina, As I recall correctly, the thing with Spark Iterators is that they use a generator to iterate through the data, so in a sense they do not keep...

Hi Christina, Ohhhhh, yes. I see now. You are right. This needs to be addressed. I'll check in with the Expedia guys because I think they are unaware of this....

Hi Christina, Thanks, let me know if you come up with something we can discuss it in this thread. I'll ask other people for their opinion. However, I the Expedia...