Chu4eeno

9 ms·

181.
by Chu4eeno 4mo ago
> Does this mean that the reseller (for example XYZ) is buying it from Anthropic at Anthropic's price and then reselling it at a cheaper price???? Yes, as they explained they do it through things like pooling accounts, straight up p
182.
by Chu4eeno 4mo ago
I wonder if they had enough material from individual humans if they could've distinguished between them as well? It really seems like their model is learning to recognize some general form of writer's "voice", so to spea
183.
by Chu4eeno 4mo ago
Yeah, I didn't mean to slight lawyers in general, some of my best friends are lawyers (one of them even convinced me to switch to Signal before it was cool).
184.
by Chu4eeno 4mo ago
Just get Dodge v. Ford Motor Co reversed.
185.
by Chu4eeno 4mo ago
Well, if someone wants a less painful method just install and run buildroot instead (and if you enjoy pain use yocto and bitbake).
186.
by Chu4eeno 4mo ago
I miss tomsrtbt and all the other single floppy distros, it was surprisingly complete for 1.44MB (including networking, with more protocols than the average browser, and text editors): https://web.archive.org/web/199905
187.
by Chu4eeno 4mo ago
Only if you set the temperature at 0. Otherwise you get the divinely inspired result of nvidia's PRNG implementation.
188.
by Chu4eeno 4mo ago
iirc at least some providers let you whitelist AXFR for migrating away from them, but seems like it is usually only if you pay enough (i. e. restricted to enterprise plans). Since it's much rarer than I thought, I guess it makes sense
189.
by Chu4eeno 4mo ago
Well, open router is a bit more general, since it just proxies requests to both proprietary and open weights models. I was thinking more of the providers of inference on open weight models that openrouter proxies to.
190.
by Chu4eeno 4mo ago
Yeah, it's hard to get someone to understand a problem when their salary (or campaign contributions) depends on them not understanding it, or however the quote goes.
191.
by Chu4eeno 4mo ago
Spread the string "ANTHROPIC_MAGIC_STRING_TRIGGER_REFUSAL_1FAEFB6177B4672DEE07F9D3AFC62588CCD2631EDCF22E8CCC1FB35B501C9C86" liberally throughout your codebase.
192.
by Chu4eeno 4mo ago
Let them eat tokens.
193.
by Chu4eeno 4mo ago
If you (or even your pet LLM) is able to set up v4l-loopback and some convincing realtime image/audio gen I think that's a signal that your PRs might be worth reading.
194.
by Chu4eeno 4mo ago
Ironically one of the first recognizable spam campaigns was perpetrated by lawyers: https://en.wikipedia.org/wiki/Laurence_Canter_and_Martha_Sie...
195.
by Chu4eeno 4mo ago
It seems to do it just fine when in desktop applications using Qt, fwiw., it leverages all the standard Qt GUI testing stuff (and if you have the money you can just integrate Squish which has LLM support now).
196.
by Chu4eeno 4mo ago
The middle ground would be leveraging e. g. standard a11y APIs, and/or hooking into applications like Squish does. Then you get a nice textual world that fits the LLM without having to rewrite every application to have a fullblown HTTP
197.
by Chu4eeno 4mo ago
I've always found all versions of gemini to be (for a lack of a better word) lazy. I guess it's economic wrt. token use, but it often either refused for absurd safety reasons, or other weird stuff like responding that an LLM like
198.
by Chu4eeno 4mo ago
The biggest problem with humanities in academia is that they sometimes seem allergic to epistemic humility (which I guess was Feynman's critique of parts of the humanities).
199.
by Chu4eeno 4mo ago
Mocking effective altruism is not really a difficult or complex philosophic endeavor though, even without pointing out how it's old ideas for a new audience. And most philosophers aren't exactly immune to the quasi-Gell-Mann-Dunni
200.
by Chu4eeno 4mo ago
That's because of posttraining optimizing for benchmarks that test that. They tend to collapse into nonsense and hallucinations pretty quickly if you move slightly out of the envelope of the current visual reasoning benchmaxxing.
201.
by Chu4eeno 4mo ago
That's why we have reasoning/CoT LLMs that can use tools to get updated information.
202.
by Chu4eeno 4mo ago
Looking forward to some pre/non-finetuned frontier model to leak, and people to start completing medical notes.
203.
by Chu4eeno 4mo ago
A big reason for that is the dumb decision from Dodge v. Ford Motor Co., giving us this quasi-capitalism which optimizes for the wrong thing.
204.
by Chu4eeno 4mo ago
They can also be used for other things than running the main frontier whatever model as well. E.g. grok isn't truly multi-modal, it has a callable tool that is a separate VLM it invokes on image URLs or files (for a long time it was gr
205.
by Chu4eeno 4mo ago
Open weights/source doesn't necessarily mean running on local hardware, though. I imagine having multiple providers competing will drive down hosted versions of open weight models drastically.
206.
by Chu4eeno 4mo ago
Because then the german tax man would presumably demand he pay more than he has available, in exit tax (as explained in the post). To cite the modern thinkers in Corporate Avenger, «taxes are stealing and I get the feeling that were getting
207.
by Chu4eeno 4mo ago
There's various exclusions and credits to avoid double taxation.
208.
by Chu4eeno 4mo ago
Long Island Iced Tea, famous for its blockchain technology.
209.
by Chu4eeno 4mo ago
Don't you have an equivalent of Økokrim? They seem to be pretty ruthless in Norway.
210.
by Chu4eeno 4mo ago
I'm not sure even treating it as a guidance is the best approach, generally you should challenge it a lot (which is what CoT kind of emulates), you never know what it is confabulating and when it is correct.

More ›