pmarreck

11 ms·

211.
by pmarreck 6mo ago
I did not know about the trivially-produced botulinum toxin potential of garlic sitting in olive oil at room temperature. I'm going to guess that asking a cloud censored/non-abliterated LLM would not get me this information, despi
212.
by pmarreck 6mo ago
The abliteration in particular that Heretic does apparently results in a best-in-class lack of "stupefying" the underlying model. You haven't read its claims, apparently.
213.
by pmarreck 6mo ago
I was not aware of this. I might not be willing to trade accuracy for speed in this case, then.
214.
by pmarreck 6mo ago
Holy hell, how new is this? I've never heard of it, looks great!
215.
by pmarreck 6mo ago
I need something like this but for cross platform utilities
216.
by pmarreck 6mo ago
That's awesome!
217.
by pmarreck 6mo ago
I'm not sure what you're angling at but I already gave a set of questions that are ethically legitimate yet routinely censored by the public models: https://news.ycombinator.com/item?id=47654013 Not to mention tha
218.
by pmarreck 6mo ago
Really? That's fascinating. Why is that?
219.
by pmarreck 6mo ago
1) Coming up with any valid criticism of Islam at all (for some reason, criticisms of Christianity or Judaism are perfectly allowed even with public models!). 2) Asking questions about sketchy things. Simply asking should not be censored. 3
220.
by pmarreck 6mo ago
Impressive model, for sure. I've been running it on my Mac, now I get to have it locally in my iPhone? I need to test this. Wait, it does agent skills and mobile actions, all local to the phone? Whaaaat? (Have to check out later! Anyon
221.
by pmarreck 6mo ago
so does insisting on TDD. I've known these things from the beginning. Any extra restriction that still produces functional code ends up being great for LLMs to curb them deterministically.
222.
by pmarreck 6mo ago
About 4?
223.
by pmarreck 6mo ago
LOL I have yet to push mine up. Suck it, Dario. Anyway, Gemma4 just came out and is pretty good and can be made to work with Openclaw (currently dealing with a timeout issue though)
224.
by pmarreck 6mo ago
I do all my dev in CC now. Only occasionally review with an editor. CC is great at many things but one area it is still not great at is making GUI interactions look and work properly. Literally likely because it can neither see the GUI (wit
225.
by pmarreck 6mo ago
Can quantum computing do even basic math yet? I think this was the holdup. Or perhaps I'm missing the point.
226.
by pmarreck 7mo ago
This one has a bit of... an Art Deco flavor, perhaps, is it?
227.
by pmarreck 7mo ago
That is usually configurable at the terminal level- for example, both wezterm and ghostty have available configs to control this behavior.
228.
by pmarreck 7mo ago
Berkeley Mono was the first time I bought a font. It's so good. Perfect even. And they have a really neat customization tool. I've been using it for a few years now and they actually still occasionally release a new version of it.
229.
by pmarreck 7mo ago
Hope they enjoy working on Java code... Forever... With 3 month release cycles, no CD... LOL
230.
by pmarreck 7mo ago
Yeah but they're not owed the telemetry. It's a privilege (if you convince your users to interface with your service via their client), not a right.
231.
by pmarreck 7mo ago
These techniques were all the rage on early Macintosh things
232.
by pmarreck 7mo ago
By that definition, every single human who wrote a paper after reading a source document is a “plagiarism machine” and I’m 53 and well remember Symbolics from freshman year at Cornell, in fact my application essay to it was about fuzzy logi
233.
by pmarreck 7mo ago
You sound like someone who has literally zero understanding as to why that is a ridiculous comparison. There are a thousand and one ways that I participate when building something with LLM assistance. Everything from ORIGINATING AN IDEA TO
234.
by pmarreck 7mo ago
Who prompted the LLM? Who vetted the output? Who ensured there was adequate test coverage? Who insisted on a certain design? Who is to blame if it's bad code? That is the same entity that is responsible, and the same entity that "
235.
by pmarreck 7mo ago
Instead of arguing for violating the boundaries of a "slow, bespoke" no-LLM project, you can simply start one that enjoys all the benefits of LLMs by NOT having that boundary. Very simple solution.
236.
by pmarreck 7mo ago
> author your contributions yourself This is such an easily refuted assertion. Tell me, if something is wrong with the submitted code, who or what is responsible? If it's not "the LLM", then your opinion makes zero sense.
237.
by pmarreck 7mo ago
This is nonsense. There's plenty of work that goes into it. In fact, if no human work goes into it, then it is unlikely to pass human muster/judgment. It is just a tool for accelerated work, like literally every technological pro
238.
by pmarreck 7mo ago
LOL, ah, whoops then =)
239.
by pmarreck 7mo ago
This is a non-argument. All of the cloud LLM's are going to move to things like micronuclear. And the scientific advances AI might enable may also help avoid downstream problems from the carbon footprint
240.
by pmarreck 7mo ago
so a perverse incentive, basically. shocker. got it

More ›