igul222

Try removing BN from the critic :)

> It is not for lack of trying that all the top papers in visual question answering end up doing this as a classification task. Results are really poor when it is used as RNN generation

I'd be curious to know if you have a reference for this. Given that the answers are one word, a word-level RNN language model output should basically be the same thing as a straight 1000-way softmax.

This is also not quite right. Models whose parameters are too big to fit on one GPU can be trained by splitting them across multiple GPUs, as was done here, for example: http://papers.nips.cc/paper/5346-sequence-to-sequence-learni...

Cool algorithm! One problem is that its probability density calculation assumes a random ship placement. If the opponent were aware of how the algorithm works (let's say they deduced it from its behavior, which seems plausible) then they could devise a counter-strategy of placing ships on the squares which have the fewest other possible ship placements (e.g. in the beginning of the game, this is along the edges of the board).

According to their "About" page, a one-way ride costs $22. Even if you buy 50 rides at once, the price is still $17.6 per one-way ride. That's comparable to what a typical UberX ride around Palo Alto costs, in my experience.

Patenting it doesn't necessarily mean they won't let other people use it. It could, for example, be for protection against Apple patenting it themselves and not letting anyone use it.

It's not a problem if those connections use self-signed certificates, right? If that's the case, then setting up SSL from CloudFlare to your servers should be pretty easy.

I don't know whether it's possible with ZIP, but there are definitely compression schemes for which this can be done in time proportional to the size of the compressed file.

More likely then not, you'd just get noise.

The article mentions that commercial napkins sold for 4 rupees despite cotton being 1/4000th the cost, but also says farther down that the locally produced ones sell for an average of 2.5 rupees each. That's a 38% discount: significant, but it doesn't seem like a game-changing difference to me. I imagine that if this catches on, the larger manufacturers will likely just cut their margins to compete.

On a more general note, one of the big reasons for the Industrial Revolution's switch to mass-production was that making goods in huge factories is more efficient, and ultimately cheaper, than producing them locally in small quantities. Economies of scale are powerful.

Guessing no, because of the Node.js plugin architecture ("need to call into C or C++?", "full access to the filesystem"). Really only the UI seems to be done in WebKit.

See CGI.

Don't yank the winners' prizes, but re-judge the entire hackathon and award the new winners equal prizes.

USV.com post about this HN post: http://www.usv.com/posts/union-square-ventures-new-website-i...

I organize a 1,000 person hackathon at MIT, and the entire organizing team is super-excited about using this. In hindsight, GoComm seems so simple and obvious that one wonders why nobody built it before, but I guess that's true of all the best products (Dropbox, Stripe, etc.)

1 is configurable – when you configure Watchsend you can pick whether you want to record everything or only record certain screens.

2, 3, 4, and 5 are in the works :)

Cofounder here. We're a service that helps app developers understand their users in order to make their apps better. Like almost any other service that collects and stores data, it's possible to intentionally misuse Watchsend in ways that grossly violate users' privacy.

What we mean by "we take privacy seriously" is that we work really hard to make sure that devs who care about their users' privacy can use Watchsend to make their apps better without infringing on their users' privacy. At the end of the day, if a developer wants to spy on their users, there's very little we (or anyone else) can do about it.

I think Heroku's default Rails app server is WEBRick: https://devcenter.heroku.com/articles/ruby#webserver

Try asking at https://www.facebook.com/events/1374512966106958/ – someone else might be in your situation.

Email team@hackmit.org and we'll help you out :)

Ishaan from HackMIT here. Regarding interest, in the past week, we've gotten over 1500 signups –– more than we can handle, to be honest. But there's definitely no lack of interest.

Whether you pay to opt out of ads is a function of how much you hate ads and how willing you are to spend money in general. I think the latter factor is more significant.

The problem is if Google were to do that, the value of their ads would plummet since the kind of people who would pay to opt out of ads are the same people who would be most likely to pay for things in ads.

Not 100% sure but I think this is is the same reason TV companies don't want to offer their content for streaming online, and Hulu can't remove ads for its premium users.

> Travel times between the US and China, or between Australia and Europe, are now comparable to those of the great age of sail in the 19th century.

If this is true, why don't we all just use sailing ships, which require no fuel and produce no emissions?

I'm not suggesting replacing the second for scientific work-- that would be a mess, I agree. But since it seems the day is already good enough for casual timekeeping, why not replace the non-scientific, imprecise, everyday uses of second/minute/hour with fractions of the day?

What makes the metric system appealing is that each physical quantity has exactly one "base" unit, and you can make easy conversions– just multiply or divide by 10.

The thing about time is, there are two important 'natural' units that aren't going away: the day and the year. Whatever system you devise has to include them somehow. And unfortunately, the conversion factor between them (365) isn't a multiple of 10.

That said, we can do a lot better than the status quo. Since we've already given up on the idea of having a single base unit for everything, why not express time in fractions of days instead of multiples of seconds? Counting time up to 1.0 days is a lot more intuitive than counting up to 8.64 myriaseconds. The second is a unit of time which isn't tied to anything intuitive (actually, it's entirely arbitrary). If we're going to shake things up, let's at least use days instead, yeah?

Has any hacky workaround like this ever held up in a court?

How's this?

  git clone https://github.com/whatever; heroku create; git push heroku master

The problem with this and most similar projects is that I probably don't want all of these defaults, and I don't have time to look through each one. In fact, on average, these defaults are probably worse than OS X's default defaults.

If someone made a simple interface to let me cherry-pick the ones I want, that could be useful.

Does anyone know what algorithm this CP implementation is using behind the scenes? I'd be a lot more impressed if it were using dynamic programming than plain brute force.