> 4x150 = 1200
o\
181 karma · joined January 26, 2016
o\
Hopefully I'm not violating any rules by sharing this thing that looks like it might be TON's leaked whitepaper
If you don't mind relocating to Chicago drop me a line at kirill.sevastyanenko (at) avant
All I want is to be able to do this: 1) specify a DAG of tasks. Each task is a docker image, CMD string, CPU and memory limits 2) hit an API to run it for me. Each task runs on a new spot instance 3) be able to query this service about the state of the DAG and of each individual node
Sounds like if AWS provides an API to create a batch cluster (or whatever you call it) and lets the tasks be defined in terms of what docker image to run with what command you'll satisfy this desire
However it's worth pointing out that theano's API is somewhat similar to tensorflow so migrating shouldn't be too hard and should be fairly easy to test
Soviet Union famously regulated this very well. These luxury goods were only available to accredited investors aka the Party elite.
It seems to me that back then this behavior was condemned by the people of the Land of the Free.
HFT is mostly ultra-fast arbitrage programs, and the benefit of using them far outweighs the downsides.
Maybe the problem is not with HFT but in the industry as a whole
Back when Yahoo! was a real company they used a technique called "multi-armed bandits" to learn what ads to show. [1]
More recently, there's a number of off-the-shelf packages available that you can trivially integrate into your ML pipeline to optimize hyperparameters of your models, I'll include the links below.
[1] multi-armed bandits https://www.theregister.co.uk/2011/09/23/yahoo_core_personal...
[2] tree of parzen estimators https://jaberg.github.io/hyperopt/
[3] hyperband - what google uses in their internal ML toolkits AFAIK https://arxiv.org/pdf/1603.06560.pdf
[4] (shameless plug) gaussian process based hyperparameter optimization service https://github.com/avantoss/loop
In terms of a new backend, there's uber's pyro [1] that's based on pytorch, and then there's Edward [2] that's based on tensorflow (and I think recently got integrated into tensorflow repo). Would be interesting to see pyMC adopt MXNet as its backend. I'm sure Amazon will become a heavy user and might sponsor the development. Or at least the authors will be able to get jobs there without a whiteboarding interview phase.
I'm talking more about the whole experience of building out a full app. How do I structure my codebase? Which packages should I use in 2017? It's a much more trivial exercise with vue if you're only doing frontend work occasionally and not as part of your normal job.
I have to point out that https://github.com/facebookincubator/create-react-app is great if you do decide to go with React and want answers to aforementioned questions.
> simpler syntax. I think single-file components is a great thing when you're getting started. While true "js ninjas" will call me out on this, they're probably employed by a large org with problems that are different from mine.
> less choice. I only write frontend occasionally. How would I know which react-router to choose? What's the best redux package? What about handling CSS, what's the flavor du jour? Vue says "want something done - do this. here's vue router, vuex, etc."
Overall, the two are very much similar in both how they work and what they provide, but one is developed and used by an organization that has a virtually infinite supply of engineers, and the other is being developed by one guy.
Technology that makes Facebook successful may be inappropriate for a successful hackathon or a small startup
And those that do, they are on forums to learn about new exciting shiny frameworks/languages, like Elixir
Alternatively Bitmain has just plugged in a new batch of their ASICBoost
if you couldn't care less and just want to be able to earn your bread - AI won't take your bread unless you're an uber driver or a customer support employee
One thing that would make playing with this tech more interesting to me and other newcomers is a guide on how to create a new environment for gym or universe, sort of a crash course on what steps need to be made in order to apply your algorithms to my existing problems
https://www.recode.net/2017/3/8/14863560/uber-ai-gary-marcus...