nharada

26 ms·

331.
Ask HN: How to accommodate everyone's office preferences?
1 points by nharada 5y ago|1 comments
332.
by nharada 5y ago
There's an interesting viewpoint that comes through in these comments that seems to imply that the policy and regulatory side of climate isn't "doing", and that engineering or science roles would be more beneficial to ma
333.
by nharada 5y ago
So this is purely for inference right? Basically you're setting up a physical analog inference device by writing weights to flash memory and then using the device properties to do that mat-muls required for the forward pass? This is su
334.
by nharada 5y ago
There's an interesting answer and linked article here: https://academia.stackexchange.com/a/2948
335.
by nharada 5y ago
I often see "we" on internal documents written by one person -- I think it comes from the convention in academic papers where you'd use "we" to refer to the researcher(s), even if it's just one author. I don&#x
336.
by nharada 5y ago
Super interesting. I grew up in a nearby suburb and when I was younger Motorola was THE company in the area. It felt like nearly all of my friend's parents were employees. It's amazing how quickly they fell off.
337.
by nharada 5y ago
Awh, I thought that meant infinite flight time, but it appears instead it carries the RTG along to recharge Li-ion batteries at night.
338.
by nharada 5y ago
This is an awesome response, thanks for so much detail!
339.
Ask HN: What Was the iPhone Release Like?
6 points by nharada 5y ago|9 comments
340.
by nharada 5y ago
Yeah, I own an EV with piss poor fast charge rates so it pains me to say this, but charging per minute is probably the right move. It certainly feels like the bottleneck right now is charger availability, not amount of electricity available
341.
by nharada 5y ago
Dang that's a brutal commute
342.
by nharada 5y ago
> it's much easier to manage junior than seniors, more so when you are not an engineer yourself I definitely have not found this to be the case. Good senior folks require very little active management.
343.
by nharada 5y ago
Latency, inference cost to Google. Privacy, in theory
344.
by nharada 5y ago
From the abstract[1]: > After decades of effort, 17% of the total residues in human protein sequences are covered by an experimentally-determined structure. Here we dramatically expand structural coverage by applying the state-of-the-art
345.
by nharada 5y ago
Thanks for the link and the blog post. Diffusion models have definitely been making waves over the last year or two and I've been slacking on really digging in.
346.
by nharada 5y ago
Feels like at a certain point those people will just go to a "be the last one standing" mindset.
347.
by nharada 5y ago
> Given that this trend is now stalling naturally, the warming trend should follow suit Even a cursory argument should make it obvious this is untrue: CO2 is cumulative, so emitting the same amount year over year will continue increasing
348.
by nharada 5y ago
The human eye has incredible dynamic range -- it's something like 25-30 stops total (although you can't see them all at once). Compare that to the 10-12 stops for a high quality DSLR and the 5-7 for the types of cameras going into
349.
by nharada 5y ago
I love the title sequence (even though that was a different design shop): https://www.artofthetitle.com/title/the-expanse/ They clearly drew from some other highly technical areas in their design: for example inco
350.
by nharada 5y ago
I'm curious what you would suggest as an alternative. Completely flat orgs have worked occasionally in the past(?) but in general I hear horror stories about those as well.
351.
by nharada 5y ago
It looks like many of the components used are still English biased, for example the off the shelf tool they used to generate phonemes from text was a state of the art system for English and a more general but less performant version for oth
352.
Biden plan could save California high-speed rail if state leaders can ever unite(politico.com)
1 points by nharada 5y ago|0 comments
353.
by nharada 5y ago
The urgency is different. You don't earthquake-proof your house while the living room is on fire.
354.
by nharada 5y ago
I don't really understand this. I just looked into moving onto EV rates for PG&E and the super cheap "off peak" hours start at midnight and run through the night. If solar is causing issues with overcapacity during the da
355.
by nharada 6y ago
You (and many others) might have this viewpoint, but it's the loud minority that causes the pain. Even with the small amount of open source software I've maintained I've gotten all kinds of angry emails and completely useless
356.
by nharada 6y ago
Curious how big of an issue this actually ends up being since they can't be put in a stable higher orbit without a second burn/collision to circularize. Without a change to the orbit perigee it would still decay, just more slowl
357.
The Affective Growth of Computer Vision [pdf](authentic.sice.indiana.edu)
2 points by nharada 6y ago|0 comments
358.
by nharada 6y ago
> The resources needed to forge a signature aren't going to be much different than the resources needed to create a deepfake to begin with I imagine deepfakes will become way easier in the near future because there are legitimate co
359.
by nharada 6y ago
I'm not sure I take your point here. Broadly, humans. More specifically, a company, a government, a country, an open source community, a family. Is your point that there be disagreement on ML policy? I don't think that makes any d
360.
by nharada 6y ago
Even if you think the example presented here isn't a big deal, it still exposes a larger issue in these large language models. These models learn from historical data and that history may not be the behavior we want in the future.

More ›