"Beware of geeks bearing open weights"
501 karma · joined December 24, 2015
Its using WebGPU as a proxy to estimate system resource. Chrome tends to leverage as much resources (Compute + Memory) as the OS makes available. Safari tends to be more efficient.
Maybe this was obvious to everyone else. But its worth re-iterating for those of us skimmers of HN :)
I just hope Google doesn't kill this one as quickly as they did Stadia etc.
Ever since mobile & cloud era at their peaks in 2012 or 2014, we’ve had Crypto, AR, VR, and now AI.
I have some pocket change bitcoin, ethereum, played around for 2 minutes on my dust-gathering Oculus & Vision Pro; but man, oh man! Am I hooked to ChatGpt or what!
It’s truly remarkably useful!
You just can’t get this type of thing in one click before.
For example, here’s my latest engineering productivity boosting query: “when using a cfg file on the cmd line what does "@" as a prefix do?”
Do you think or worry about not-being able to test these things? (Or is that just me :))
Details: I ack/understand this comes from a dependency (ReAct agents); not directly langmanus.
But, still, curious what the community/hn-tech thinks of testability, veracity, potentially conflicting or overlapping instructions across agents, etc, wrt “prompts” as sources of logic. Ack its a general practice with LLMs.
Running either Grok2 or DeepSeek or even Llama405b requires nearly 400-500gb of memory.
Buying a tinybox with enough gpu memory costs $15k-25k. Or equivalently the same if you build your own.
A distributed Mac cluster costs about the same, if not more, if you’re buying 2-3 M2 Ultra each with 192gb of memory.
So people are absolutely constrained by price/supply here. Every engineer, analyst, scientist would be far more untethered by rules & regulations or policies & terms-of-service nitty gritties if they can trust that LLM they use is completely local, without-telemetry or tracking and is licensed fairly for commercial use (perhaps this excludes llama).
Not a lot of people can afford $15k-30k in spending for a computer (that can run this sota llms). But you can a billion will buy one when it’s $1k
Despite all the layoffs & black founding fathers debacles Google as an institution has had recently, it still has the systems in place to let passionate engineering projects see the light of day.
That’s really cool!
Is there a necessity to always keep a valid visa stamp in my passport? Apart from the ease of travel are there any other reasons to always keep a current visa stamp in my passport? I do have all valid I-797 documents of status and have kept my status current all the time.
Nobody cares about anything. Somebody cares about something. Everybody cares about everything.
I think the truth is somewhere in the middle. This makes me optimistic. At least, its nice that people care to hold doors open : )
is that really a deal-breaking decision to buy an iMac? Yeah, sure, I agree it's super silly design that they put the port under the mouse; but c'mon, does it really matter?
As far as I can tell, I have never had to explain to my 61 year old Indian mother how to use a Mac as much as I have had to debug every little thing on Windows PCs. Macs & Apple products _truly_ do "just work"
Doesn’t matter what the computer becomes — AI, AGI or God-incarnate — there’s always a role between that and the end-user. That role today is called software engineer. Tomorrow, it’ll be whatever whatever. Perhaps paid the same or less or more. Doesn’t matter.
There’s always an intermediary to deal with the shit.
Hmm, I wonder if that’s the roles priests & the clergy have been playing all this while. Except, maybe humanity is the shit God (as an end user) has to deal with
I set up a whole dotfiles tracking system and periodically keep making backups of the config. Fish has been sole system that has travelled with me to different machines, different companies, running on Macs, running on Linux, running on personal laptops, iMacs, and so so many different versions/instances of workstations on the server in docker, in aws etc etc.
I worked at Amazon in 2017 and we had a whole "developer environment" system built off of Apollo and my Fish config fit right in with the company-defaults for various build systems, log systems, metrics yada yada. I wrote myself a ton of nice aliases, new functions, new scripts.
I moved to Apple in 2019 and brought my Fish config over (sans any Amazon-specific things of course) and all my customizations have ported over nicely. They play well with all the Apple-y unique configs this company now gives its engineers.
Fish has been snappy & delightful.
I guess one could argue zsh could work just as well and has the benefit of being compatible with native bash syntax. But I found zsh and its ilk (oh my zsh) too slow tbh. They're nothing --and I mean zilch, nada-- compared to the speed of Fish. Fish w/ Bass (love the puns in the fish community, btw) accomplishes much of the backward compatibility needs with bash or bash-like syntax; while still performing at the snappy speeds of Fish.
Love Fish! Love Julia's notes! Ahh life is just perfect sometimes!
While we're fanboying over Julia, here's a picture of the time I made myself a t-shirt from Julia's zines of Recurse Center's values: https://x.com/b0rk/status/876571293491109889
Well if your argument is: “a perfectly good 10 year old iPad is held hostage by Apple because they don’t provide side loading” then isn’t that the same case with your perfectly good 10 year old Toyota? Or 10 year old Target Toaster oven or 10 year old Samsung fridge?
I think we will live in a strange world if we start demanding our refrigerators allow side loading YouTube.
There’s hardly any order of dipping, if any.
Why doesn’t the web compete away Apple’s “dips”?
But if there's anything he's proven it's that he can ship product. Tesla cars are phenomenal. xAI looks exactly like what Tesla looked like in 2010. Far too futuristic & optimistic timelines that peers & laypersons would characterize as fluffy grandiose snake-oil.
But I bet they will ship. And it won't be fluffy nor snake-oil.
1. No error message. Just recover the cached message.
2. The message is deleted
There’s a world we live in where ahem-ahem a certain design focused company would’ve done it this way.
Ultimately, it’s the responsibility of the Mail app to make a decision on whether to recover the cached message or not. It’s those engineers & designers who built that damn cache in the first place. And the server too.
Let _them_ deal with the consequences of messages existing in one but not the other. It’s not the user’s job to care.
This definitely seems like a Google building products people will love. Let’s just hope they don’t kill it in 2 days/weeks/months/years
Todays ML frameworks grapple with the problem of “jacobian-vector products” & “vector-jacobian product” as a consequence of understanding the interplay between gradients & derivative; and the application of the “chain rule”. All of those 3 concepts are fundamentally understood by being proficient in calculus.
While I’m being the hype-man for calculus I don’t mean to say proficiency in linear algebra or statistics is in any “less necessary” or “less useful” or “less challenging” or “less..” in any way.
I’m merely stating that, historically, calculus has been the unique branch of study for engineering. Statistics has always found value in many fields — business, finance, government policy etc.
Sure Linear algebra is one of those unique fields too — I kinda like to think of it as “algebra” in general and perhaps its utility has flowed in tandem with calculus. Idk. I haven’t thought super hard about it.
My strong “opinion” here comes from an observation that while calculus may have been a required subject of study in awarding engineering degrees, the reality is, people didn’t really study it. They just brushed through a couple of pages and wrote a few tests/exams.
In America there’s plethora of expert software engineers who opt for a bachelors degree in computer science that is a BA not a BS.
I think that’s complete totally reasonable thing to do if you don’t want to grind out the physics and calculus courses. They are super hard after all. And let’s face it, all of the _useful to humanity_ work in software doesn’t require expertise in physics or calculus, at least until now.
With AI going forward it’s hard to say. If more of the jobs shift over to model building then yes perhaps a back to basics approach of calculus proficiency could be required.
It was not really necessary through all of the app developers eras. In fact, it’s so much so the case that many software engineers graduating from 2000-2015 or so work as software engineers without a degree in BS. Rather, they could drop the physics & calculus grind and opt for a BA in computer science. They then went on to become proficient software engineers in the industry.
It’s only after the recent advances of AI around 2012/2015 did a proficiency in calculus become crucial to software engineering again.
I mean, there’s a whole rabbit hole of knowledge on the reason why ML frameworks deal with calculating vector-Jacobian or Jacobian-vector products. Appreciating that and their relation to gradient is necessary to design & debug frameworks like PyTorch or MLX.
Sure, I will concede that a sans-calculus training (BA in Computer Science) can still be sufficiently useful to working as an ML engineer in data analytics, api/services/framework design, infrastructure, systems engineering, and perhaps even inference engineering. But I bet all those people will need to be proficient in calculus the more they have to deal with debugging models.
I don’t mean to disregard other branches of math — of course they’re useful — but calculus stands out in specific _applicability_ to engineering.
Literally every single branch of engineering. All o then. Petrochemical engineering to Biotech. They all use calculus as a fundamental block of study.
Discovering new drugs using Pk/Pd modeling is driven by modeling then drug<->pathogen repo as cycles using Lotka models.
Im not saying engineers dont need to learn stats or arithmetic. IMO those are more fundamental to _all_ fields, janitors or physicians or any field really. But calculus is fundamental to engineering alone.
Perhaps, a begrudging exception I can make is its applications in Finance.
But every other field where people build rockets, cars, airplanes, drugs, or ai robots, you’d need proficiency in calculus just as much as you’d need proficiency in writing or proficiency in arithmetic.
There was brief sidestep from late 1980s to early 2010s (~2012) where the term "software engineer" came into vogue and completely ran orthogonal to "proficiency in calculus". I mean, literally 99% of software engineers never learned calculus!
But it's nice to see that ever since ~2015 or so (and perhaps even going forward) proficiency in calculus is rising to the fore. We call those "software engineers" "ML Engineers" nowadays, ehh fine by me. And all those "computers" are not people anymore -- looks like carefully arranged sand (silicon) in metal took over.
I wonder if it's just a matter of time before the carefully-arranged-sand-in-metal form factor will take over the "engineer" role too. One of those Tesla/Figure robots becomes "proficient at calculus" and "proficient at calculations" better than "people".
Reference: [1]: I took the terms "engineer" and "computer" literally out of the movie "Hidden Figures" https://en.wikipedia.org/wiki/Hidden_Figures#Plot
It looks like ever since humankind learned calculus there was an enormous benefit to applying it in the engineering of rockets, aeroplanes, bridges, houses, and eventually "the careful arrangement of sand (silicon)". Literally every one of those jobs required learning calculus at school and applying calculus at work.
"No, we should be use cheap commodity abundantly available cpus and orchestrate then behind cloud magic to write our nl translation apps"
or maybe "no we should build purpose built high performance computing hardware to write our nl translation apps"
Or perhaps in the early 70s "is anyone seriously considering personal computer hardware to ...". "no, we should just buy IBM mainframes ..."
I don't know. Im probably super biased. I like the idea of all this training work breaking the shackles of cloud/mainframe/servers/off-end-user-device and migrating to run on peoples devices. It feels "democratic".
Nobody is entitled to dictate what _software_ should be on the device. If end-users have the right to _software_ then that right cannot be selectively applied in the case of phone or laptops: it should fairly applied in all cases: including your refrigerator. Why doesn’t the government compel Samsung to enable Spotify & third party App Stores on those devices?