tulio_ribeiro

6 ms·

1.
by tulio_ribeiro 18d ago
Good outcome, mostly. Option E beat literally every other option, so it is an unambiguous Condorcet winner. Bradley–Davidson analysis: https://gist.github.com/tulior/74479b28ddd1e191e5b2d89b8b68c... The Bradley-Davidso
2.
by tulio_ribeiro 19d ago
Great insight. Let me see what ChatGPT can cook if we classify each. The tally is now available; might be useful. ### Edit Re-classified using permit, prohibit, materially_restrict or none_of_the_above as I think these are more accurate lab
3.
by tulio_ribeiro 19d ago
It is. Thanks for the update.
4.
by tulio_ribeiro 19d ago
All this for an SDK you might not need. SDK is just a wrapper around POST https://api.openai.com/v1/responses Just REST + JSON. streaming (sse parsing) + tool calling + structured outputs + pagination are easy to do.
5.
Debian has published the official results for the 2026 GR on LLM usage(vote.debian.org)
14 points by tulio_ribeiro 19d ago|7 comments
6.
by tulio_ribeiro 19d ago
Good outcome, mostly. Option E beat literally every other option, so it is an unambiguous Condorcet winner. Options: https://www.debian.org/vote/2026/vote_002 The tally.txt file is still unavailable. Bradley–David
7.
by tulio_ribeiro 1mo ago
In the article, she credits the 2023 DeepMind paper Language Modeling Is Compression as the source. She also links to this post from 2015: https://colah.github.io/posts/2015-09-Visual-Information/
8.
by tulio_ribeiro 1mo ago
This one is also very interesting: https://programbench.com/
9.
by tulio_ribeiro 2mo ago
The lifetime rendering-time savings wouldn’t even offset the five minutes required to install it and make it my default terminal. I would need to display roughly 70 GB of benchmark-equivalent ASCII output before Shitty recovers a five-minut
10.
by tulio_ribeiro 2mo ago
I’d go even further and add that these distinct tools don’t belong in the same toolbox.
11.
by tulio_ribeiro 2mo ago
This is obligatory read: https://blog.royalsloth.eu/posts/it-takes-a-phd-to-develop-t... Later, Microsoft fixes the issue but fails to give Muratori credit. After backlash, they went back and gave him a footnote: http
12.
by tulio_ribeiro 2mo ago
I fail to see the problem in that.
13.
by tulio_ribeiro 2mo ago
As the other comment stated: different purposes. Still! I appreciate your sharing this. I’ll try this later today.
14.
by tulio_ribeiro 2mo ago
Hey, man. Great work and thanks for sharing this as CC.
15.
by tulio_ribeiro 2mo ago
The whole anti-LLM crusade feels like developers trying to gatekeep their own relevance. They know the tool can automate parts if not all of what gave them status, so instead of adapting, they want to declare its use illegitimate and keep t
16.
by tulio_ribeiro 2mo ago
Yes, they are. But that’s exactly what Proposal B (the only one without moral panic) says: allow its use, but place the responsibility on whoever uploads it. If you upload broken or unlicensed code, that’s on you. Proposal B is the most pra
17.
by tulio_ribeiro 2mo ago
Supervised learning and reinforcement learning are mathematically distinct. Supervised pre-training as you said maximizes the likelihood of a token sequence. RL optimizes a policy against an external reward signal or execution environment.
18.
by tulio_ribeiro 2mo ago
Needs to be a member of parent #codeberg, which is invite only. I wonder if they gated the room after public backlash?
19.
by tulio_ribeiro 2mo ago
I’d just like to add, and this may interest you both, that I’m not disagreeing with either of you. When 3.5 Flash first showed up in AI Studio, its model card said it had a March 2026 knowledge cutoff. After the backlash over it apparently
20.
by tulio_ribeiro 2mo ago
Weird choice of SSIM/RMSE. By feeding these back into the model, the agent is actively degrading its artistic output. This is only valid for the target reference I think? Alternatively, I think much better results could be had by compu
21.
by tulio_ribeiro 2mo ago
Didn't know a rocky planet in the habitable zone of a red dwarf could retain atmosphere against intense stellar stripping. Red dwarfs are known to be cooler (the habitable zone is therefore closer) and unstable. I don't think LHS
22.
by tulio_ribeiro 2mo ago
Thanks for doing the work and hosting this. This will very quickly become one of my daily drivers.
23.
by tulio_ribeiro 3mo ago
HN clients are one of the most popular projects. I’m gonna agree with some of the other comment: this is nothing new. https://hnpwa.com/
24.
by tulio_ribeiro 3mo ago
I don’t think the Firebase HN API has rate limits. By the way, how many concurrent requests are you doing?
25.
by tulio_ribeiro 3mo ago
LG’s guidelines state: - “Your app should only collect the minimum user data required for providing service and should avoid collecting unnecessary data.” - “LG performs security reviews on submitted apps before distribution, using the vuln
26.
by tulio_ribeiro 3mo ago
Cloudflare shows a 59/41 split: https://radar.cloudflare.com/adoption-and-usage?dateStart=20... HE shows 41% ASNs support v6: https://ipv6.he.net/
27.
by tulio_ribeiro 4mo ago
Maybe these are what you mean? https://github.com/jdevalk/specification.website/blob/main/p... https://github.com/jdevalk/specification.website/blob/main/m...
28.
by tulio_ribeiro 4mo ago
What’s up with case 09058169? Seems like a 5 minute fix
29.
by tulio_ribeiro 1y ago
"I, a notorious villain, was invited for what I was half sure was my long-due comeuppance." -- Best opening line of a technical blog post I've read all year. The narrator's interjections were a great touch. It's rar
30.
by tulio_ribeiro 2y ago
Sweeteners are processed food. Timeline shows more processed food hitting the market, period. Obesity rises. Coincidence? Doubt it. It's not just the sweetener itself. It's the whole shift. More processed crap in everythi

More ›