jorgemf

53 ms·

541.
by jorgemf 10y ago
> If you work really hard, learn all of this stuff, and do a really good job and save your company a bunch of trouble by knowing all of this, you won't get a dime for it, and that's the problem I think that is your problem. I d
542.
by jorgemf 10y ago
The research you have to do is based on your previous knowledge. The less you know the more you need to learn when you are facing a problem you don't know how to solve it. But people is lazy, if they know a way to solve it they will go
543.
by jorgemf 10y ago
I was not talking about implementing anything. I disagree with you in a lot of things. I don't like whiteboards because they are stressful and feels strange if you don't use them regularly. I was a teacher which uses whiteboards d
544.
by jorgemf 10y ago
It is not false, for example Albert Einstein was developing his Theory and he could do it because he was in a research ground where they were doing research about infinitesimal calculus, which was handy for Einstein to apply to his theory.
545.
by jorgemf 10y ago
Yes, this is what I meant
546.
by jorgemf 10y ago
With this type of post I always read comments complaining why they should know those things for an interview. They are not necessary. It is like some more knowledge is going to harm them. I wish I would know more about everything. Sometimes
547.
by jorgemf 10y ago
> "before you ask me to demonstrate a depth-first search, you had better explain to me why I'm going to need to be doing that instead of just writing an SQL query and tweaking an index" How I am going to ask you to tweak a
548.
by jorgemf 10y ago
As an engineer you should be able to see the global picture and know other things. Because you wont be able to use something to solve your problem if you don't know it in advance. I mean, you don't need to know the details, but yo
549.
by jorgemf 10y ago
Yes, but I know it is going to be very difficult to find someone willing to pay as much as I expect for the time I spent developing it (1 year full time + 1 year part time). So, it will be always mine until I decided its time to kill it for
550.
by jorgemf 10y ago
Buy an Nvidia GPU with 8Gb of ram and you are good to go by now. Or you can use amazon instances or google cloud. For toy things I wouldn't bother to suffer the hell that is to compile it (I do it for my job and it is a pain to suffer
551.
by jorgemf 10y ago
It is not Google's fault. You need to agree the terms and conditions of Nvidia in order to download cuDNN. So you have to do it manually.
552.
by jorgemf 10y ago
> But it's a bit disappointing to hear that the build system is something of a nightmare I know... I am building it for TensorFlow Serving and it is pain in the ass. I will help but I have no idea how to do it, only find workarounds
553.
by jorgemf 10y ago
Those issues are related with TensorFlow <1.0. It wasn't an stable release. Try 1.0.1.
554.
by jorgemf 10y ago
>The standard build uses a "least common denominator" Intel instruction set (SSE4) TensorFlow is not meant for the CPU. If you want to do something serious you have to move to the GPU, which it is at least 8 times faster than y
555.
by jorgemf 10y ago
Only for the models, the core works with python 3. I tweaked the models to use them in python 3, it is things like 'xrange and range'.
556.
by jorgemf 10y ago
It is not a good idea to compile TensorFlow by your own unless you really need it (for example for TensorFlow serving). Python packages are the way to go.
557.
by jorgemf 10y ago
> This article is based on Your Language Sucks in the form of half a joke. First I though the article was being ironic, I end up understanding by thinking it is a completely a joke, not half a joke. Otherwise I do not understand: - why t
558.
by jorgemf 10y ago
I was wondering the time per step, I am aware it converges faster. Let me know when you have more results! I want to start doing things with text soon and I would love to try RWA
559.
by jorgemf 10y ago
Sounds promising! A couple of questions after a quick read, but sure I have more later: Is it faster than LSTM for training / inference? and, are you planning to use it for more complex problems as translation or image to caption?
560.
Ask HN: What would you want the HN community know more about or be aware of?
7 points by jorgemf 10y ago|4 comments
561.
by jorgemf 10y ago
They should ask you whether you want just to join the company in any role or you want to be working in a specific role.
562.
by jorgemf 10y ago
I did an interview with Google few years ago, I still don't know the position they wanted me for. I make pretty clear my preferences but I was interviewed by random engineers not related with what I wanted to do in Google. I guess they
563.
by jorgemf 10y ago
I don't want to sound negative and I might be wrong in my assumptions but 50-75k sounds like something you can get from FFF (family, friends and fools) or supply it with your time. So the question for me is why an investor is going to
564.
by jorgemf 10y ago
actually you do all of them, all of them help to overfitting
565.
by jorgemf 10y ago
There are a lot of solutions for the problems you mention. For overfitting you can do data augmentation, normalization, dropout and early stop with the test set. (and probably improve your dataset) More neurons means more parameters to adju
566.
by jorgemf 10y ago
You can not upgrade, but you will have security issues. It is very hard to offer support for legacy systems. You should expect more than 2 year support. Software evolves really fast, companies cannot sit and wait. They will lose their hype
567.
Principles to guide the development of Artificial Intelligence by researches(futureoflife.org)
2 points by jorgemf 10y ago|0 comments
568.
by jorgemf 10y ago
I don't understand your comment. I am saying you should have updated SO and updated apps. Apple is in charge to update the SO and the app developers to update their apps.
569.
by jorgemf 10y ago
Why do you care about security fixes when updating the OS but you don't care about security fixes of apps? If an app is 2 years old it is probably using libraries with security issues. I would blame the developer for not updating the a
570.
by jorgemf 10y ago
I know, but doesn't make sense to offer them in a couple of years. When we already have TensorFlow running in GPUs in amazon or other companies. Actually we made numbers in my company and Amazon GPUs are cheaper than Google Cloud with

More ›