imh

79 ms·

751.
by imh 11y ago
Wouldn't this require larger datasets? That isn't always an option. I'm imagining that a smaller, more computationally efficient network could learn nearly as well with fewer data points given these heavily engineered feature
752.
by imh 11y ago
I think NLP is really cool, but it seems to be moving so quickly. If I wanted to get a decent overview, are there some review papers or textbooks with good coverage that aren't too out of date?
753.
by imh 11y ago
Not only is it rude, but it's not a social gathering in the same way. That's how I read this quote from the article, >After five days without phones or tablets, these campers were able to read facial emotions and correctly iden
754.
by imh 11y ago
This makes it impossible to do blind review (review without knowing who the author was), which seems desirable. Do existing math journals do blind review?
755.
by imh 11y ago
Yup. That's been my finding as well. char-rnn was really just a diversion of curiosity after I'd cleaned up the data. My best idea right now is to make a generative model of p(next_token | previous_token(s), author), essentially c
756.
by imh 11y ago
I'm actually running that right now with HN comments. It's not done training, but it's not that much more interesting than OP's. Here's some example output: {"text": "The article was in client 70, or
757.
by imh 11y ago
I think I can give pretty good insight into this, as someone with a weird background coming to programming. I'm a data scientist with a physics background, but I'm still even atypical in that role, having just a bachelors degree.
758.
by imh 11y ago
>I don't see any posts where authors concern themselves with metaphors for Maxwell's equations (or anything else in physics; maybe I'm hanging out in the wrong circles though). This is quantum mechanics through and through
759.
by imh 11y ago
Man, I really would have expected scientific american to do a better job with their units. Power (watts) is a rate, so a 100 kW facility will power 120 homes, not "120 homes for a year."
760.
by imh 11y ago
Can we start getting a label/tag for sites that hijack scrolling? The way we label pdfs and articles from previous years?
761.
by imh 11y ago
Here's the next article in the series: http://slpopejoy.github.io/posts/Effectful02.html
762.
by imh 11y ago
I wonder how much it is related to the fact that we hardly know the people around us. I don't know my neighbors. I've heard that's a new phenomenon. We, the adults, might not even step out of our bubbles as much these days, m
763.
by imh 11y ago
I've been something wondering something for a while now. Rust is supposed to be a kind of C/C++ replacement, right? But the last I heard it's still much slower. So what's the roadmap to speeding it up? Googling for it fa
764.
by imh 11y ago
> you can't train motivation. That's where that pesky paycheck comes in.
765.
by imh 11y ago
>So why do we keep doing it? Why can’t we put the book down? So that I can keep my job. Easy.
766.
by imh 11y ago
>...VCs treated their LPs a bit like mushrooms, keeping them mostly in the dark... The full phrase is to treat your <insert business relation> like mushrooms: feed them shit and keep them in the dark.
767.
by imh 11y ago
I go to a thai place in SF that skirts the ban by using the heavy "reusable" bags as disposable ones. There's no difference between these and actual disposable bags other than that they are thicker. They're still small a
768.
by imh 11y ago
Most journalists choose to present a simplistic view of the world. Greenwald seems unafraid of saying "the truth is complicated," and then proceeding to tell a readable, worthwhile story. Every time one of these comes up, I make a
769.
by imh 11y ago
Ah, I couldn't tell how you meant it. Poe's law and all that.
770.
by imh 11y ago
Does that make twitter, amazon, and tesla still startups? I swear I'm not cherrypicking data. Those, plus google and facebook, are the first big companies I thought to look up P/E on. Public startup seems like a contradiction. ht
771.
by imh 11y ago
Not Agile enough :p
772.
by imh 11y ago
Can we please stop calling billion dollar companies startups?
773.
by imh 11y ago
Jesus christ, is the argument here, "You're all paid too much and don't deserve it. Let's push you harder?" There's a huge difference between pushing ourselves and what we're hearing about Amazon.
774.
by imh 11y ago
Other good versions of this concept: https://hbr.org/2014/12/vision-statement-how-to-lie-with-cha... http://data.heapanalytics.com/how-to-lie-with-data-visualiza... There's an amazing version
775.
by imh 11y ago
FYI this page looks really bad on firefox with a narrow window. It gets jumbled around and the dark blue background on the left takes over the whole page.
776.
by imh 11y ago
> Burnout seems to almost always affect founders whose startups are not doing well... There doesn't seem to be an obvious causation here. I could totally see either causing the other, or there being a common cause of both. Seems wei
777.
by imh 11y ago
I'd love more stories like this. It's weird to think about some of the big gaps in knowledge I take for granted. There's plenty we learned through school about the world around us, but a surprisingly small amount of that came
778.
by imh 11y ago
People always say this about physical tech, but the difference is ease and scalability. Slashing all the car tires in a block is harder and more traceable than sending out a small RF signal.
779.
by imh 11y ago
At the same time, if I was an early Uber employee and the difference between current value and exercise price is in the millions of dollars, then I'm on the hook for a tax bill I can't afford. Are there any solutions beyond taking
780.
by imh 11y ago
The point is the drone will be gone by the time the police arrive, so there is no drone to follow.

More ›