Serious number crunching on Google Cloud
March 14, 2017 — July 30, 2018
computers are awful
concurrency hell
premature optimization
I want to do cloud machine learning using Google’s CloudML offering.
Google cloud might interoperate well with a bunch of Google products, such as TensorFlow. I use that. Let’s see how it works.
1 Notes
Obviously, this will work better if you commit to Google storage APIs, etc.
Note (2017-02-14): Google supports Python 2.7 only — so be prepared to party like it’s 2010. (Is this still so?)
Seems to also support VMs?
There is the usual thicket of weird service names.
Basic workflow bits are
- cloud storage stores data
- dataflow supplies data
- model fitting done using CloudML
- control this from Datalab
Also good to know:
2 Painstaking run through
See How is Google number crunching awful?.
Google now has their own instruction manual for one common use case, TensorFlow. TensorFlow without a PhD.