suprjami

7 ms·

1.
by suprjami 4d ago
"If only there was someone could stop me, the CEO of Torment Nexus Inc, from creating the Torment Nexus."
2.
by suprjami 5d ago
I skimmed chad the other day. There's very little to it (by design). afaics you should be able to replicate what chad does by copying the chad system prompt into a SYSTEM.md for Pi. Then you could use it with whatever model/provid
3.
by suprjami 5d ago
> Coding is solved, perhaps, with unlimited token spend on a frontier model. Coding is also a solved problem with unlimited salary budget on the very best developers.
4.
by suprjami 5d ago
Qwen 3.6 27B and 3.8 27B are the darlings of local inference at the moment. The only other thing anyone is using is Qwen 3.8 Flash Next, only by memory-rich people. Depending on which benchmarks you believe, these models (and the Ornith 1.5
5.
by suprjami 8d ago
Updated the rootless podman container and restarted, worked fine.
6.
by suprjami 9d ago
Strange comment. "If you remove the hype about how transformative they are, they really are transformative". No. If you remove the hype about how transformative they are, you're left with what they actually are: a sometimes m
7.
by suprjami 11d ago
Yes, all the people talking about this benchmark since OP created it.
8.
by suprjami 11d ago
> The Astra pelicans are much better. It's absolutely in the training data now. Time to retire this (previously mildly amusing) benchmark as frontier models are now pelicanmaxxed.
9.
by suprjami 11d ago
This just lists every possible configuration permutation. Nothing insightful here.
10.
by suprjami 15d ago
So this is like OpenSCAD (code-based 3D modelling) but for music?
11.
by suprjami 17d ago
I didn't immediately get the reference. He is the Korean Go player who lost to AlphaGo in 2016.
12.
by suprjami 17d ago
Please don't forget you are under no obligation to triage all the awful things. You are well within your rights to require submitters reach a certain standard before you will review their request. Your time is finite and precious. Your
13.
by suprjami 18d ago
AND you made it for Sega Genesis! That's so good.
14.
by suprjami 18d ago
People used to say the same about Linux ext4, it didn't need defragmenting. afaics all that meant was file allocations make a little extra space at the end, so if the file expands it doesn't need to allocate a new block. If one ke
15.
by suprjami 18d ago
The NAS of Theseus
16.
by suprjami 20d ago
That makes more sense now. Thank you for the explanation!
17.
by suprjami 21d ago
This is even more baffling because Source Sans is open under SIL which permits free commercial use without restriction. So if the goal was to save money, wouldn't the smart move be to use Source Sans for $0, instead of paying the desig
18.
by suprjami 25d ago
Now someone vibe-code a browser extension which detects this and removes it. A uBlock Origin rule for the whole domain of anybody who does this would be fine too.
19.
by suprjami 26d ago
I feel you have both missed the point of caveman reasoning. LLMs are trained on human natural language, not a specialised internal-only monologue to make syntactic shortcuts. Their response should make grammatical sense to a human reader be
20.
by suprjami 26d ago
Another William Gibson story.
21.
by suprjami 26d ago
And from John's parent comment, "The future is here, it's just not evenly distributed." is a quote attributed to William Gibson. We're all just quoting authors now.
22.
by suprjami 27d ago
This feels like a lot of words to say "a decimal point between two integers".
23.
by suprjami 1mo ago
So-called "caveman" thinking attempts to address this. The important part of "actually wait, I really need to XYZ" is just "XYZ". The model can attend to just "do XYZ" and produce almost the same vect
24.
by suprjami 1mo ago
Thanks, today I learned!
25.
by suprjami 1mo ago
> and why it is not enough on its own Nothing is good enough on its own. Geoblocking, fail2ban, port obscurity, SSH keys, limiting logins to specific usernames, not using your public internet nickname, putting things behind CloudFlare tu
26.
by suprjami 1mo ago
You might find this article relevant: https://news.ycombinator.com/item?id=49151933
27.
by suprjami 1mo ago
Top-K: example setting 20. Select only from the 20 most likely tokens. Top-P: example setting 0.9. Select tokens whose probably accumulates to this number. So say you have tokens with 0.7 then 0.2 then 0.1, the last will not be selected bec
28.
by suprjami 1mo ago
Temperature, top-up, top-k, min-p all control which token the model predicts next and how likely it is to select one token over the other. You might understand this as "The capital of France is..." and the model isn't always
29.
by suprjami 1mo ago
You have understood correctly. One really would think these companies (including Google) who spend many millions of dollars on compute could write a few hundred lines of Jinja correctly, so their investment works optimally or at all. But th
30.
by suprjami 1mo ago
Our desire for better local models just happens to coincide with China's desire to destroy the western AI company business model by releasing local models. I doubt there's any philanthropy involved.

More ›