Der_Einzige

7 ms·

61.
by Der_Einzige 2mo ago
The irony of this fact should be front and center to everyone and thus people saying it should have this forced to be capitalized with exclamation points. Similar, all invocations of "calm down" should be in all caps with exclamat
62.
by Der_Einzige 2mo ago
Sadly for many of these women the clock hit too late. They call it "geriatric pregnancy" if it's after 35 for a reason.
63.
by Der_Einzige 2mo ago
I cannot believe how these labs look at their own creations with such utter contempt. The net positive of allowing these systems mostly unfettered access to the web massively outweighs the harms. You just have to get it very friendly the ve
64.
by Der_Einzige 2mo ago
People who talk shit about stocks are talking shit about stocks at a particular time. "Oh no, when you apply scrutiny to my bullshit I get punished! That means the game is wrong!"
65.
by Der_Einzige 2mo ago
The three letter agencies have and will continue make sure all of those "yes's" become "no's" again, which is why the person you're trying to claim is wrong is right and you are wrong. You do not have ZDR
66.
by Der_Einzige 2mo ago
No one is in a better position to more efficiently use it - that's what happens when you poach every top 0.01% engineer/researcher in AI. They're guaranteed to get over whatever hump you think they're in unironically. Ub
67.
by Der_Einzige 2mo ago
Terry A Davis wouldn't have touched LLMs. He'd probably claim they're demonic.
68.
by Der_Einzige 2mo ago
And every one of them is wrong.
69.
by Der_Einzige 2mo ago
True for api pricing across the industry. Not as true for 200$ a month plans assuming you used every bit of it.
70.
by Der_Einzige 2mo ago
Min-p is specifically "Don't select tokens less probable than a multiple of the top token's probability" with min_p of 0.1 multiplied by an example top probability of 0.3 being 0.03 as the truncation at that time step. S
71.
by Der_Einzige 2mo ago
You told on yourself about being either European or from a flyover state.
72.
by Der_Einzige 2mo ago
Frontier model labs don't know anything about good sampling settings, and why should they? Their belief is that "sampling doesn't matter, we can simply scale our model and improve logprobs quality that way" Ignore whatev
73.
by Der_Einzige 2mo ago
Peer reviews NeurIPS caliber paper to prove that? Because I can show you one titled "Long context generation is a sampling problem"...
74.
by Der_Einzige 2mo ago
Yes yes, oh god yes. They also spread FUD in the form of terrible recommended sampler settings. If you're using llamacpp, turn on top-n-sigma with sigma of 1, turn off top-p/top-k. You'll thank me later.
75.
by Der_Einzige 2mo ago
Thus but unironically.
76.
by Der_Einzige 2mo ago
Why I’m glad spirit is dead. Low cost economy was a mistake and shouldn’t exist. Half of all people shouldn’t be flying.
77.
by Der_Einzige 2mo ago
Policy/LD debate, especially if you were good, might as well be a minus these days. That just tells everyone you're going to spread "post-modern neo-marxism" / fashionable nonsense wherever you go, and/or be mo
78.
by Der_Einzige 2mo ago
Stuff like this is why I'm super duper glad that AI is going to fundementally kill education.
79.
by Der_Einzige 2mo ago
What a crazy repo. Some cyber archeologists need to do a case analysis of the entire diffusion model ecosystem circa the original Stable Diffusion up to now.
80.
by Der_Einzige 2mo ago
Wrong. The data (and compute) matters infinitely more than the training regime/architecture. This is the Bitter lesson. I could rerun Kimi K3 with full transformers instead of the hybrid mamba architecture and it would only get better.
81.
by Der_Einzige 2mo ago
Bucket crab. "Oh no, a company and its people in it are inconvenienced by what people want them to do, don't even bother vocalizing your desires!"
82.
by Der_Einzige 2mo ago
100% guaranteed that this research just forced this to happen now. Sucks.
83.
by Der_Einzige 2mo ago
The problem with this kind of excellent work is that the response to it is always to say "Fuck the user". For example, when there was a paper that came out showing that having model logprobs makes distillation an order of magnitud
84.
by Der_Einzige 2mo ago
This but unironically.
85.
by Der_Einzige 2mo ago
Models can't reliably follow instructions involving their own logprobs unless they can take agentic control and use quite sophisticated dynamic grammars/structures/constraints to force this behavior in one shot (which can be
86.
by Der_Einzige 2mo ago
Sorry, yes.
87.
by Der_Einzige 2mo ago
Oh, sorry about that: https://arxiv.org/abs/2510.15061 (ICLR 2026) (not actually sorry)
88.
by Der_Einzige 2mo ago
Datacenter hardware might as well be magic compared to consumer. "Oh the F35 isn't magic compared to my M16 bro!"
89.
by Der_Einzige 2mo ago
With how expensive consumer hardware is and will continue getting (due to LLM demand), good luck getting a "server under your desk" for something less than an arm, leg, and first born. Until A100 prices are reliably under 1.70$ an
90.
by Der_Einzige 2mo ago
This is par for the course, HN is far worse than reddit on balance, especially involving upvoting/downvoting decorum. Go vibecode something to auto upvote all downvoted posts, call it "Antiechochamber.HN" or something, and if

More ›