trvz

5 ms·

1.
by trvz 3d ago
1) usable for most tasks != pleasent to use for most tasks 2) using an Intel Mac as your main computer, for most tasks != using it as a home server
2.
by trvz 3d ago
And given the performance jump from Intel to Apple Silicon, the perceived difference is 10+ years even.
3.
by trvz 4d ago
Don't mind this, they'll definitely be fished to extinction now.
4.
by trvz 10d ago
Indeed, and because of that, I argue it's not just the crime of hacking, but crimes against humanity, due to endangering the existence of the species.
5.
by trvz 11d ago
It's a bigger model, of course it's slower.
6.
by trvz 13d ago
Normal people: tok/s or t/s Psychopaths: tok/SEC
7.
by trvz 13d ago
Why? Sure, I’d prefer it, too, but this is just another GNU/Linux vs. macOS situation: most of us would prefer the first, but actually get shit done on the latter.
8.
by trvz 13d ago
Simply use GLM 5.3 Flash on the Strix Halo you've purchased just for this situation.
9.
by trvz 14d ago
Just put Mythos on the task; it’ll work out the best way in a measly few hours.
10.
by trvz 14d ago
Try turning the sound on, off, on again — not impressed by this bugginess.
11.
by trvz 14d ago
As a European: I don’t care where an open weight model comes from.
12.
by trvz 16d ago
I don't disagree with the title of this weblog post, but its premise is wrong: the existence of AI has an impact on the total number of tech companies, but not on the number of companies that make it real big (the AirBnBs, Stripes and
13.
by trvz 16d ago
For most people employed as programmers, inheriting code from previous people has been a thing. Also relevant: including libraries written by others in your own software. As always it comes down to finding a healthy balance.
14.
by trvz 20d ago
You misunderstood what I was referring to by “hacker” there.
15.
by trvz 20d ago
First, smaller models are fun for hackers: you can run them locally, or run them faster. Second, when cloud models become unavailable or otherwise deteriorate, these will be all you have. May as well prepare.
16.
by trvz 21d ago
All M3 variants use LPDDR5.
17.
by trvz 21d ago
That M3 had an older type of RAM. Apple hopefully secured sufficient supply of the newer variant for the M5 Ultra.
18.
by trvz 21d ago
It’s implied that they do, but don’t have the balls to tell you they do.
19.
by trvz 23d ago
"matches" could've also been written as "this year's product falls short of the competitor's last year one". And multithreaded is 10 cores vs 6 cores. I appreciate Apple being kept on their feet, but they&
20.
by trvz 23d ago
Strix Halo is 256 GB/s.
21.
by trvz 24d ago
Simple, ask the AI to give you a prompt that triggers extensive research or work into a topic.
22.
by trvz 25d ago
You used the wrong units there. Should be Mbps and Gbps.
23.
by trvz 27d ago
It’s not even greed or corruption – it’s mental illness.
24.
by trvz 28d ago
I think the appropriate way in this case is to cover it up with black tape.
25.
by trvz 1mo ago
> How this was written: drafted with an AI assistant from my own notes, then edited by me. Chill, dude. The AI datacenters generate enough heat.
26.
by trvz 1mo ago
Literally the same security risk.
27.
by trvz 1mo ago
“High parameter […] models” was mentioned above. It’s still not a good take however, as laws of civilizations getting more convoluted over time is the general behaviour, not specific to LLM-producing countries.
28.
by trvz 1mo ago
Agentic behaviour isn't just coding. In fact: > We recommend using it for agentic workloads, tool use, data extraction, RAG, and long-context workflows. It is not recommended for agentic coding and knowledge-heavy tasks.
29.
by trvz 1mo ago
No. No.
30.
by trvz 1mo ago
They're not. When things go wrong it's better to compromise someone else's VM host than your own computer. It's only a matter of time now until AI will find novel ways to break out of virtualisation.

More ›