- We've built a local vector database with every SEC filing over the last few years. And we've built a tool call on top of that to allow these LLMs to read and query sec filings.
- Have done the same for a lot of other data sources, just giving the LLM access to them and allowing it to spend some time to actually research.
Just to cut through the headline here. The largest chunk of Oracle layoffs were in India [1]. In comparison, they've barely fired any American workers.
Contrary to popular opinion, IT workers aren't interchangeable and there exist knowledge a large swath of jobs that very few people qualify for because of the specialization required.
America is at near full employment [2]. Replacing American workers with lower paid foreign workers is already illegal and frequently enforced[3].
This is such a deep distraction but a virulent virus of a narrative.
For tech jobs specifically? Compensation has been increasing since the turn of the millennium, what standard of living do you mean? If you mean housing, that's due mainly to NIMBYism from native labor buying and owning houses, especially before the tech boom, not imported labor.
There are lots of APIs with poor or nonexistent documentation. I'm talking about internal systems where one programmer that kinda knew what he was doing built a proof of concept, and now it's a core business requirement.
Cheap labour producing goods for the native population at low costs should increase your standard of living, no? It makes the products you buy cheaper.
By your logic, if you were the only person in the country, you'd live like a king.
I've spent US$16 700 last month. I made an autoscaling K8s cluster for distributed compilation on a large C++ project. I also heavily modified the build system to use a forked version of `siso` compatible with our environment.
That meant we can go from 17 minutes on 32 cores to 5 minutes on a few hundred. And because it's distributed compilation we don't have to provision each developer with an overpowered build system they won't be using most of the time.
It could also eliminate our CI backlog because autoscaling. Over a few hundred engineers building this codebase this probably a few thousand hours of waiting a week.
This took me about 2 weeks as someone who graduated 9 months ago. Most of the tokens were spent in several hour long debugging sessions relating to distributed systems networking and tracing through gRPC logs because the system wasn't working until it did.
I think I'd need several years of experience and 6 months as a full time engineer to have accomplished the same thing pre-AI.
Since I work at a semiconductor company near Toronto there's nobody around with the distributed systems experience to mentor me. I did it mostly on my own as a side project because I read a blog post. I literally wouldn't have been able to complete this without AI.
I'm sure the actual solution is terrible compared to what a senior developer with experience would've created. But my company feels like they're getting ROI on the token spend so far even though it's double my salary.
The orthogonal LoRA constraint is interesting. Have you thought about whether orthogonality conflict with the timestamped training? If two temporally adjacent observations should produce similar LoRA updates, orthogonality would actively push them apart. Maybe you want similarity for recency, orthogonality only for distinct episode types?
How is this whataboutism? Saying that the idiot in charge of the DoD isn't going to change a long standing hostility towards human life from the Iranian regime is whataboutism in what way? Please indulge me. You're genuinely expecting me and everyone to just forget the 70s hostage crisis, the state sponsorship of terrorism, and the recent slaughter of 40k civilians and with a straight face say if they kill a US soldier it's because of...Pete Hegseth at a press conference?
Many sellers will cut whatever corners they can to get a lower price point, as that's what purchasers look for. The one that stands out to me is shipping, sure go for for cheap shipping on a trivial cost item, but I question doing the same when you're buying something expensive and not consider spending some proportion of the price on a better courier/service tier to have more certainty the item will get to you and in good condition, assuming the seller doesn't bake-in the cost of upgraded courier.
PostSscript files came from the same company in the 80's
-Adobe- and with GhostScript and zmachine.ps you can play Zork I-III, Calypso, Tristam Island and the rest of propietary Infocom text adventures on it with ease.
What zmachine.ps does is to emulate the ZMachine VM in PostScript and display the output to stdout/console.
And if it werent for the PS stack limitations for sure you could emulate Linux under Risc-V.
Heck, you can emulate old RISC Linux' syscalls (enough to run static binaries) in Perl without ever calling to C bindings, not even once.
Thanks for sharing this. I love Merlin but never knew how they got it to be so good. Blood, sweat, and tears - of course - as everything actually valuable and useful requires.
Hey, I'm only 3 months into attempts to clearly uninstall MSI software form my laptop, so i can reinstall it (it refuses to install atm), because it seems to be the only way to...... set the battery charging limit.
They couldn't do that for "a few bucks of nano banana credits" though. You could generate the imagery but that's only one line of evidence. A launch is easily detectable through multiple signals.
Why would Russia and China and any other country with any degree of astronomic capability that the US has an adversarial relationship with just let them get away with lying to the world? Why wouldn't they take the opportunity to humiliate the US by revealing that no launch happened and that they cannot detect the spacecraft?
