tough

11 ms·

181.
by tough 1y ago
I love both Go and Rust but Go is GC no? Where rust shines is on its safety at compile time and lifetime/borrow/memory model imho both are awesome tools!
182.
by tough 1y ago
yeah they had an outage in 2024 i think or was that registrations only?
183.
by tough 1y ago
how i solved for this was reciprocal pinning so any node owns all their data (personal log) + whoever they follow (mutual follow reprocity) last x content (indices so its easy to get last n, or last since xyz) and so anyone can decide exact
184.
by tough 1y ago
how i would make it work is you'd only download actual bits you care about, the rest can be handled by indexers you pay for to help you sort it out
185.
by tough 1y ago
I actually tried to vibe-code this, not finished tho
186.
by tough 1y ago
Nice read > I'd be grateful for any pointers to an example where system developers (or someone else in a position to know) have verified the success of a prompt extraction. You can try this yourself with any open source llm setup th
187.
by tough 1y ago
This is aawesome, thanks for the links and the write ups!
188.
by tough 1y ago
i think you are pretty good yourself joohnisgood
189.
by tough 1y ago
if you're in a regulated field like law or medicine and you fuck up signing some AI slop with your name, you should loose your license at the very least tools are fine to use, personal responsability is still required. Companies alrea
190.
by tough 1y ago
karma please go up
191.
by tough 1y ago
yeah i call mine secure-runner lmao go binary
192.
by tough 1y ago
someone on twitter was exploring and linked to some related papers where you can for example trim experts on a MoE model if you're 100% sure they're never active for your specific task what the bigger wide net bigs you is generali
193.
by tough 1y ago
both opencode and charm's crush support LSP's and MCP's as configs
194.
by tough 1y ago
> Is it still satisfying if that software is bad, or harms many of those people? sadly many engineers might detach or feel powerless or simply not their part of the job, and only be attracted to the huge scale. sadly you can only ever wo
195.
by tough 1y ago
if anything i would do differently, i'd try things only machines can reliably do. unless the llm and the design for it is necessarily adversarial, not even going into red teaming or jailbreaks. A human couldn't type for 24h straig
196.
by tough 1y ago
afaik now FAIR lives under the MSL umbrella https://x.com/AndrewCurran_/status/1940109829781270799
197.
by tough 1y ago
Both OpenAI and Anthropic do the out-of-band to a certain degree, the only issue is until now sycophancy has been a feature not a bug (better engagement/retaining cohorts) so go figure
198.
by tough 1y ago
The new Facebook AI Czar wang hinted on previous interviews that Meta might change their stand on licensing/open source. Seems like the tides are shifting at meta
199.
by tough 1y ago
that relies on the positive-aligned RLHF models most labs do. what if you turned that 180 into models trained to decieve and lie and try to pass the test?
200.
by tough 1y ago
yes that's the benevolent take,but in a market for attention, it feels like a bit of a vampire attack to a brand to spun off OpenBrand. seems some what of a template now, OpenCut was recently on HN because some chinese company already
201.
by tough 1y ago
Zuck has promised allegiance to trump, so what's new here
202.
by tough 1y ago
zuck hired also dana the ufc guy no? not like he seems to give a fuck
203.
by tough 1y ago
yes, weird, lol
204.
by tough 1y ago
im gonna bet this was gpt-5 as patio11 mentioned 1 line suggestion > It’s a little rich, pardon the pun, to cash a check for $20 billion and then whine about fintechs freeriding on your IT spend.
205.
by tough 1y ago
maybe the best optimal strategy is leech into the brand while you're unknown, and switch to your own once you make it. A bit sleazy but it works eh Can't fight marketing and human psychology
206.
by tough 1y ago
which begs the question is it just good old EEE?
207.
by tough 1y ago
cerebras code (both sub and api) have it
208.
by tough 1y ago
The key insight here is that `comptime unreachable` isn't relying on optimizations at all—it's a fundamental part of Zig's lazy evaluation semantics. Unlike dead code elimination (which is an optimization pass that removes pr
209.
by tough 1y ago
you cannot enforce law globally online there's no internet police
210.
by tough 1y ago
ChatGPT is a sub based model, their api pricing is based on usage prob different incentives at each

More ›