jdiff

5 ms·

91.
by jdiff 2mo ago
Be nice if I could correct the specs when detected incorrectly, or check the specs of a different machine without needing to physically pull this up on each machine. Model choices all seem quite out of date. Phi-3, GPT-4o? Claude 3.5? And I
92.
by jdiff 2mo ago
One person is not a nation.
93.
by jdiff 2mo ago
Even as an informal conclusion I think it's important to know who you're up against and what their motivations are. Waving that away is how we get people like known, obvious charlatans like RFK Jr in any sort of position of power.
94.
by jdiff 2mo ago
I was objecting more narrowly to the notion that the driving force behind the obnoxious safeguards that are making the product useless was the current admin, rather than having been in there since launch and driven internally.
95.
by jdiff 2mo ago
That reasoning didn't stop them from releasing this batch of models, admittedly that may be less face to lose away from the flagship position.
96.
by jdiff 2mo ago
Weren't these obnoxious safeguards in place from Day 1, before political meddling in the most holy Free Market took place?
97.
by jdiff 2mo ago
This statement of a lack of community is at odds with the plentiful pull requests that were open before the rewrite. This is, as you say, unprecedented. Not impossible, cautious optimism is a fine response. But to bring back a point I made
98.
by jdiff 2mo ago
I use oMLX and I'm tentatively going to be giving this a shot. oMLX keeps driving me up a wall with odd papercuts, bugs, and silent failures and fallbacks that are only visible buried deep inside logs when they should be announced out
99.
by jdiff 2mo ago
No, that's not correct. That is the discussion started by the article, and the one continued in many other comment threads, even including this one far further upthread. Conversations drift over time, and this thread has evolved into a
100.
by jdiff 2mo ago
I think there's a lot of spare GPU compute. I don't think it'd be made available for this, or if it was, I believe it'd most likely be from a company that won't survive the AI bust. But I think there's a lot of
101.
by jdiff 2mo ago
Whole classes of bugs are fixed by safe rust. This is not safe rust. As a line-by-line translation with unsafe forced anywhere necessary to make it compile, this should result in no classes of bugs fixed by moving to rust.
102.
by jdiff 2mo ago
Are you interested in the discussion that was taking place before you popped in or are you just here to get outraged that something you clearly identify strongly with is getting criticism for making poor choices? "Because the discussio
103.
by jdiff 2mo ago
Bun is riddled with bugs. That's what moving fast and breaking things gets you. It gets you famously bad codebases that are nightmares to work with and slow you down over time as your sins catch up with you. Vibecoding is not enough.
104.
by jdiff 2mo ago
You don't need to look up the entire multiplication in a single move. When we do multiplication outside of our own mental lookup tables, we have an algorithm for that. The one described in the article. You can size your lookup table ho
105.
by jdiff 2mo ago
I do know how git works and in my experience, while the version number sometimes doesn't get bumped until just before a release is tagged, the version number comes from the codebase. If they're building something that says 1.4.0,
106.
by jdiff 2mo ago
That commit (and main) still purports to be 1.3.14, the version numbers haven't changed. Does that not still suggest Anthropic is using something else?
107.
by jdiff 2mo ago
> they did put the bun 1.4.0 version bump in the changelogs for claude code a while ago, over a month almost. Where? Is see no CHANGELOG in the root. I do see LATEST, last modified 2 months ago, and it says 1.3.14
108.
by jdiff 2mo ago
That PR is the rust rewrite. That's some time in the past now. As of this comment, there is no tagged 1.4 release. This is the present. Anthropic is using apparently a version of Bun that is not publicly available. This is orthogonal t
109.
by jdiff 2mo ago
There were very many open pull requests from the community that were all invalidated overnight for something that same community was assured was merely an experiment. The normal process would be "hey, this experiment is going well, we&
110.
by jdiff 2mo ago
Similar, but with better typography and responsiveness. I like actual code being present here but the majority of the challenge is just reading anything, especially when zooming does absolutely nothing.
111.
by jdiff 2mo ago
This site doesn't. Many don't.
112.
by jdiff 2mo ago
No? There are many things you are not allowed to say when advertising, many ways in which you are not allowed to advertise. Fraud is not legal, and I have yet to see anyone make a free speech case for it.
113.
by jdiff 2mo ago
> The highscores don't show the run's duration, aka the error margin Different durations have different high score lists.
114.
by jdiff 2mo ago
No, Tangled's ATProto relationship and its social features are just the only differentiating factors. They mirrored their repo somewhere else. An additional location does not make it less friendly.
115.
by jdiff 2mo ago
Project's still on Github, it hasn't migrated or anything.
116.
by jdiff 2mo ago
They're still on GitHub. This isn't any sort of official transition. It's just now also on this other ATProto-based forge.
117.
by jdiff 2mo ago
Yes, a pox on othersite, all hail currentsite.
118.
by jdiff 2mo ago
I severely doubt the world ever gets to such a point that the entire world melts into AI hallucination. And token consumption depends on so many other things, it's not all that deterministic either.
119.
by jdiff 2mo ago
The entire directory (not repo) and all repo history regardless of secrets or instructions otherwise is not needed for faster cache access. I'm not going to say that this was intentional exfiltration. Sneaky motivations are not needed
120.
by jdiff 2mo ago
It certainly is, but what does OpenAI have to do with Grok Build?

More ›