thecity2

This is like every OK Cupid profile I see in SF.

They are market makers.

That's a little sad. I used LightTable for a little while, and thought it was very cool. Was looking forward to Eve, but it was beginning to seem like vaporware, so I guess I can't say I'm surprised by this announcement.

And heat!

Not having cash actually is a feature for some establishments. There's a bar in Berkeley that doesn't allow you to pay with cash, so they had no need for registers. We assumed the purpose was to prevent any incentive to rob the place (even by employees).

You can get cheap dim sum on Clement. Half the reason I moved to the Inner Richmond. lol

You didn't hear they re-opened Sam Wo in a different location? I've been once, it doesn't have the same charm as the old place, but it's alright. Still cheap!

http://www.sfchronicle.com/restaurants/diningout/article/Res...

Unfortunately, that wasn't a joke.

Sure, but then I wouldn't have clicked.

Medium and it's not close.

You'll probably have to pay around $10 to add cellular for the watch, similar to how it works for the iPad plan.

I'm not buying one until Spotify has an app for it. I'm not switching to Apple Music just to get a watch.

This is the sense that I got. I am a one-person data science team for my startup, and I basically cobbled together most of the automation described in Michaelangelo over the course of a few months. Spinning off Spark ML jobs on EMR and saving metadata to a database.

Spark is good if it fits your use case (or if you can make it fit your use case). We have definitely found that scaling can be a real issue though.

This doesn't seem to be an open source project, unless I'm missing something. It's kind of a show-and-tell basically.

This sounds a bit like Idris with its dependent types.

I like the idea of category-oriented programming. Would be interested to see a white paper on what that means.

Unless you have like 64 cores, I don't think you will be a GPU. And yes, you can use multiple GPUs in parallel to train.

Agreed. NVDA is probably very concerned right now.

> It would probably be cheaper to just buy my own Xeon computers.

What does Xeon have to do with training? You need a fast GPU (or apparently TPU). The CPU is relatively unimportant. Furthermore, you can roll your own algorithms (i.e. architecture) in TensorFlow. You don't have to use "theirs".

Can we prove this post is incomplete though?

I'm 41 and am hoping to be engaged very soon (god and my gf willing!). I sure as hell don't expect to earn a raise. lol

It's not hard to learn a new language but it is hard to port all the old libraries to that language.

Have you heard of Mochi?

"Mochi is a dynamically typed programming language for functional programming and actor-style programming. Its interpreter is written in Python3. The interpreter translates a program written in Mochi to Python3's AST / bytecode."

https://github.com/i2y/mochi

If all that can be done in Python 3, surely it could be ported to run on BEAM to good effect.

I buy books faster than I can read them, so I get it.

I wish someone would port Python to BEAM. I'm actually surprised there isn't a project out there, except some really old stale project called beam.py I found on github, which hasn't been worked on since 2010.

Yeah, but I can see how it makes great headlines and helps them get grant money!

I'm so pissed at your ex. What an a-hole.

"20 newsgroups" is a pretty ubiquitous human-generated data set for testing LDA and other NLP techniques. I've run LDA on it myself and it recovers topics fairly well.

The thing is, reviews, messages, comments, etc, in general, do tend to revolve around some central topic(s). For example, this comment right now is about LDA. It's not totally random.

Where it's never gotten past 1955.