nmfisher

41 ms·

541.
by nmfisher 4y ago
Hit the nail on the head. Promoting housing as a vehicle for wealth accumulation will be viewed as one of the greatest failures in modern economic policy in the West. It's easy to blame Wall Street, corporations, immigrants, but the re
542.
by nmfisher 4y ago
Similarly, China is notoriously terrible at international football, despite an enormous population and it being a reasonably popular sport.
543.
by nmfisher 4y ago
Working with a spectrogram is definitely similar to working with an image, and it's interesting to think why that's the case. Take convolutional models, for example. Very effective for working with images because they're (a)
544.
by nmfisher 4y ago
ML inference in the browser seems like the most immediate use case.
545.
by nmfisher 4y ago
To me, "monetization" means everything except for just selling your product for an upfront sticker price. When it comes to games, that means subscriptions, ads or microtransactions. Arguably, the latter two aren't inherentl
546.
by nmfisher 4y ago
Depends what it is. I've had reasonable success with "validating" ASR transcripts by loading up the annotator's transcript, running the audio at 2x speed and just clicking "yes" or "no" to keep the go
547.
by nmfisher 4y ago
The blog post highlights this specific point - "US English" and "Indian English" really aren't the same English (in fact, I'd probably go even further and state that "Reddit English" and "US Engl
548.
by nmfisher 4y ago
Anyone who's dealt with any kind of human-annotated datasets would be familiar with these kind of errors. It's hard enough to get good clean labels from motivated, native-English speaking annotators. Farm it out to low-paid non-na
549.
by nmfisher 4y ago
It's worth north of $20B to Twitter shareholders and they have an (almost) rock-solid case. You can bet your ass they're going to spend the next few years to fight for it.
550.
by nmfisher 4y ago
Can someone explain what DTC actually means in this context? Is it a company selling a physical product direct to consumer via digital platforms (as opposed to "traditional" distribution channels like bricks-and-mortar retail, dep
551.
by nmfisher 4y ago
I felt the same way, but interesting that he grew up in South Korea. From what I gather, it's very similar to the Chinese system over there, particularly when it comes to maths. Endless rote learning and an overwhelming focus on standa
552.
by nmfisher 4y ago
I've only touched OpenGL very briefly, but isn't this mostly explained by the fact that CUDA is for general purpose computing and not hardware rendering? If you're mostly dealing with vertex/texture buffers, then presuma
553.
by nmfisher 4y ago
Swift on iOS still has support for C/C++ interop via bridging headers. I imagine that would be preserved even if Obj-C was deprecated as a broader iOS runtime. Eliminating C/C++ in one fell swoop would be disastrous.
554.
by nmfisher 4y ago
I don't understand - you went on keto, it worked, then you starved yourself and felt rubbish, so you gave up keto? Why didn't you continue keto without starving yourself?
555.
by nmfisher 4y ago
I totally agree, this reaction seems very hypocritical. If some rinky dink startup did exactly the same thing - as they are entitled to do under the licences of huge swathes of code on GitHub - hardly anyone would bat an eyelid. But just be
556.
by nmfisher 4y ago
That's pretty much what SLIDE [0] does. The driver was achieving performance parity with GPUs for CPU training, but presumably the same could apply to running inference on models too large to load into consumer GPU memory. https:&#x2F
557.
by nmfisher 4y ago
My surgeon brother told me that many older surgeons can't "afford" to retire, often because they lost loads of money in investment schemes (and probably coupled with an unwillingness to accept a cutback in lifestyle). You pro
558.
by nmfisher 4y ago
I dropped Thunderbird when my inbox reached a certain size. I can't remember exactly how many messages (probably 100k+), but not so many that I'd expect the whole thing to just grind to a halt.
559.
by nmfisher 4y ago
Someone else has already posted this above, but the K2 project [0] is intended to do exactly this (along with some nice integration into PyTorch). There's a big cross-over with the Kaldi team, so this seems (partly) motivated by a desi
560.
by nmfisher 4y ago
That was the idea behind ONNX, which has been somewhat successful. But if you’ve ever had to deploy these models, you’ll know it’s never as simple as promised. Different libraries use different ops, have weird export requirements or even no
561.
by nmfisher 4y ago
It's exactly what it sounds like - a criminal prosecution pursued by a private person. This is possible because otherwise, you are at the whim of the Crown (or in the USA, the District Attorney) as to whether or not a crime gets prosec
562.
by nmfisher 4y ago
Forget about writing code, simply having the ability to compile would be a huge improvement. It's obscene that I have a $1500 iPad Pro with hardware/software perfectly capable of building iOS apps, but Apple demands that I buy y
563.
by nmfisher 4y ago
For me, Spain or Italy probably competing for top spot, with one of Carrefour/Mercadona/Esselunga being best for standard (non high-end) weekly fresh shopping. I even rate Vanguard and Carrefour in China, and Fairprice/Cold S
564.
by nmfisher 4y ago
I can't tell if you're joking or not, because the quality of fresh fruit, vegetables, meat and poultry at Australian supermarkets is incredibly mediocre (and even that's being generous). I'd rank Australia near-bottom of
565.
by nmfisher 4y ago
Was this Richard Stallman’s first job?
566.
by nmfisher 4y ago
I've been tussling with my wife about putting our kid into daycare when he's 6 months old. I feel it's way too early, partly due to concerns about contracting a virus (though from the perspective that it would cause more hass
567.
by nmfisher 4y ago
I’m putting down my pitchfork, this makes all the difference. That’s totally reasonable to stop people abusing a free service (to the detriment of everyone else). Same basket as banning crypto mining as far as I’m concerned.
568.
by nmfisher 4y ago
There has been mindboggling investment in machine learning over the last few decades, yet almost nothing is protected by patent (even where patents exist, they're purely defensive). Tesla's patents were voluntarily neutered, so
569.
by nmfisher 4y ago
I meant commercialization potential for companies like Google, where anything less than a hundred million is probably a failure :) Hopefully for the non-Googles of the world (i.e. you), there's a good pathway forward!
570.
by nmfisher 4y ago
Two reasons: 1) Even though it's all technically very impressive, so far there's not a huge amount of commercialization potential here. OpenAI is charging for its GPT-3 model but its revenue is probably negligible next to the hard

More ›