julien_c

Congratulations! I'm curious, how have you saved 550k at age 32?

Is it a matter a linearly saving ~68k USD a year every working year (I'm assuming a high-paying engineer job in the Valley), or have you had equity in a successful venture?

I think this makes total sense: Bot engines (i.e., GUIs where you devise complex conversation scenarios) don't really work beyond the quick proof-of-concept project. Their quality tend to actually decrease with the amount of effort you put into training and designing the conversations.

The underlying NLP tasks however are crucial. What Wit is doing here is build on their strength on NLP tasks on a message level, and leave the conversation rules to the developer to code as they see fit.

Post-mortem would be appreciated. As someone who has Chrome extensions in the store, how did you get malware-infected?

Do you have any pictures? I'd be interested in seeing them.

Well, the stock Camera and Photos apps don't ask for permissions either.

Responses to a specific message are deterministic in this model (no diversity promoting objective implemented) so this is to be expected :)

Hey guys.

Don’t expect accurate answers here - it’s an experimental weekend hack.

With this experiment, we wanted to understand how we could train a neural network on new external datasets very quickly and study the difference in the language and answers depending on the topic of the dataset.

No scripted chatbot here, we are talking of neural-nets trained in the wild :) We are using the celebrated seq-2-seq model which computes a “thought-vector” from an input sentence and generates an output sentence conditioned on this vector. We gathered various datasets from stackexchange and launched a big overnight training of our models to have some surreal morning coffee talks with our AI.

There’s still a ton of work for the answers to start to make sense (longer training and bigger datasets would improve the quality of course, we can also easily add components to improve the variety and coherence of the responses) but a difference can definitely be noticed based on the dataset subset.

This will allow us to test a variety of new datasets way faster than before. Let us know what you think, and if you think of a cool dataset to test, just let us know on Github.

Last spring's session's videos (so 9 months ago) are up on Youtube: https://www.youtube.com/playlist?list=PLmImxx8Char9Ig0ZHSyTq...

+1

PS: Hey victor :)

Love it as well and I am recommending it to everyone I know who's shipping iOS apps

Not sure what you're saying here. They're embedding Youtube videos which YT is perfectly fine with.

You have a whitelist of possible domains though (Messaging and payments being the most common for now)

I worked on this law!

You can also just boil your pasta in the kettle.

See Github thread here: https://github.com/glock45/swifter/issues/60#issuecomment-16...

And the actual Swift package there: https://github.com/julien-c/example-package-dealer/blob/mast...

Nice job. I don't play much but this looks pretty addictive. Which tech do you use, and how do you plan to monetize?

So I'll ask the question: do this mean we're getting anywhere close to getting Swift to compile to other backends?

Hi! Thanks for sharing.

I'm looking into implementing Peek and Pop into my app as well. One thing that was not clear from your article was, do you use two different View Controllers for Peek and Pop (as the Peek VC is way simpler, though functionally related to the Pop one)?

From the Apple documentation this looked like it was meant to be one and the same VC (though displayed in a "responsive" way).

FYI, I've asked the same question on the dev forum here: https://forums.developer.apple.com/thread/21954

Cheers!

If I'm not mistaken Germany shipped nuclear waste to Normandy, France (by train) a couple of times.

It didn't fly well with either public opinions, though.

You could have a look at AIML and something like Program AB.

From https://news.ycombinator.com/item?id=10124308:

  Number of companies we offered to fund yesterday for the first YC Fellowship: 32

That's a pretty steep cut-off from the number of interviews (~120).

  Number of companies we offered to fund yesterday for the first YC Fellowship: 32

That's a figure I was looking for (took the interview yesterday, didn't get in).

Pretty cool homepage. Seems like the product (physical device + app) has a lot of features, maybe too many to get them right at the same time (I like the auto-combination of footage from multiple cameras though. But as I said, quite hard to do right).

What are the tech specs compared to a GoPro Hero 4 Session? (By the way GoPro has announced they will enter the video editing market soon).

This is really interesting, but I'll second the intuition that building a critical mass for StreetView-like data is probably one of the hardest thing to crowdsource.

It's just not fun to be driving each and every road around where you live. Besides, without an hemispheric camera data is really limited.

What would be so bad about it? Founders already committed to other projects? Leakage of "Ideas"?

Anywhere from $170k to 250k I would say.

See also the original email transcript in this other thread: https://news.ycombinator.com/item?id=9946692

Nice work. I think the git log example is contrived though, because you would probably include the real emoji characters in the actual commit title, right?

On the Apple Appstore, I've seen an average of around 30% App Store Views to App Units (i.e. downloads) being typical.

Note that those figures are historically limited because before the launch of iTunes Connect's App Analytics there was no way of knowing your number of views.

Hope this helps!

They're not shutting down as a company, though. Presumably they've burned through only half of those $4 million and will have enough air to execute something significant.