Hacker News: JustFinishedBSG

New comment by JustFinishedBSG in "Show HN: Graft – Claude Code hooks that cut grep tokens by 42%"

JustFinishedBSG — Fri, 14 Aug 2026 16:49:39 +0000

> almost as if it was trained to spit out as complex of sentences as it can.

It probably was, at least inadvertently. Very easy to imagine that one of the post-training step (RLHF, DPO etc) reinforced the "sounds clever" behaviour.

New comment by JustFinishedBSG in "The session you cannot take with you"

JustFinishedBSG — Fri, 31 Jul 2026 10:20:41 +0000

This doesn't (and cannot) solve what the post is talking about

New comment by JustFinishedBSG in "GPT-5.6 used a prompt to close a 30-year gap in convex optimization"

JustFinishedBSG — Sat, 18 Jul 2026 14:25:31 +0000

At a very high level mathematics is basically 100% text/symbolic rewriting. You start from some set of postulate assumed true and you do your thing to get a new different set of equivalent assertions in a form that is more useful.

I don’t know if LLMs will kill the working-mathematicians but at least seem like that it doesn’t seem absurd to imagine LLMs will be good at math…

New comment by JustFinishedBSG in "GPT-5.6 used a prompt to close a 30-year gap in convex optimization"

JustFinishedBSG — Sat, 18 Jul 2026 14:14:54 +0000

My experience may not be entirely representative because to be entirely honest I’m not exactly a great researcher and there are brilliant PhD students. That said it indeed was my experience that in the pre-PhD / early PhD period ( or even longer … ) your advisor proposes (gives) you pretty low hanging stuff that he mostly already knows how to solve, at least at a high level, with the expectation that it will teach you to use the mathematical tools you need.

New comment by JustFinishedBSG in "Policy on the AI Exponential"

JustFinishedBSG — Wed, 10 Jun 2026 20:08:20 +0000

It's very hard those days to think of companies/people more arrogant than Anthropic/Dario, which is quite the achievement as the bar is very high.

If that arrogance was well placed at least you could somewhat excuse it, but the fact that it is so overtly hypocritical and based on false premises just makes it so much worse.

New comment by JustFinishedBSG in "Claude Opus 4.7"

JustFinishedBSG — Thu, 16 Apr 2026 17:25:47 +0000

Interesting, it doesn't seem intuitive at all to me.

My (wrong?) understanding was that there was a positive correlation between how "good" a tokenizer is in terms of compression and the downstream model performance. Guess not.

New comment by JustFinishedBSG in "France to ditch Windows for Linux to reduce reliance on US tech"

JustFinishedBSG — Fri, 10 Apr 2026 12:54:32 +0000

Most of the cost (to the government) for Windows is "support" (in a very general sense) and that cost isn't disappearing with Linux.

Especially since it is easier to find badly underpaid (and not particularly competent) Windows sysadmins than it is to find badly underpaid Linux admins.

New comment by JustFinishedBSG in "If AI writes code, should the session be part of the commit?"

JustFinishedBSG — Mon, 02 Mar 2026 11:30:03 +0000

> for the same reason math problems require one to show their working.

We don't put our transitional proofs in papers, only the final best one we have. So that analogy doesn't work.

For every proof in a paper there is probably 100 non-working / ugly sketches or just snippets of proofs that exist somewhere in a notebook or erased on a blackboard.

New comment by JustFinishedBSG in "If AI writes code, should the session be part of the commit?"

JustFinishedBSG — Mon, 02 Mar 2026 11:26:54 +0000

> While it's noisy and complicated for humans to read through, this session info is primarily for future AI to read and use as additional input for their tasks.

Context rot is very much a thing. May still be for future agents. Dumping tens/hundreds of thousand of trash tokens into context very much worsen the performance of the agent

New comment by JustFinishedBSG in "If AI writes code, should the session be part of the commit?"

JustFinishedBSG — Mon, 02 Mar 2026 11:23:42 +0000

I understand the idea but the way I work, a commit isn't "a" session, it's potentially tens of sessions with branching in each session.

I honestly don't know if I'm doing something very wrong or if I have a very different working style than many people, but for me "just give the prompt/session" isn't a possibility because there isn't one.

I'm probably incredibly inefficient, because even when I don't use AI it is the same, a single commit is usually many different working states / ideas / branches of things I tried and explored that have been amended / squashed.

New comment by JustFinishedBSG in "Just-bash: Bash for Agents"

JustFinishedBSG — Thu, 26 Feb 2026 19:52:59 +0000

What problem were you trying to solve ? ( not that you need to solve one. I’m just curious )

New comment by JustFinishedBSG in "An ARM Homelab Server, or a Minisforum MS-R1 Review"

JustFinishedBSG — Fri, 20 Feb 2026 05:57:56 +0000

https://github.com/edk2-porting/edk2-rk3588

New comment by JustFinishedBSG in "€5k Swiss cameras with zero electronics"

JustFinishedBSG — Fri, 23 Jan 2026 18:11:25 +0000

Well according to the website you cannot buy a fully mechanical lens anymore so it doesn’t appear to be true that they sell cameras with zero electronic anymore

New comment by JustFinishedBSG in "Raspberry Pi's New AI Hat Adds 8GB of RAM for Local LLMs"

JustFinishedBSG — Thu, 15 Jan 2026 13:37:30 +0000

It's useless for LLMs and it's actually slower than Hailo 8H for standard vision tasks, so, why ?

New comment by JustFinishedBSG in "AWS raises GPU prices 15% on a Saturday, hopes you weren't paying attention"

JustFinishedBSG — Tue, 06 Jan 2026 12:54:34 +0000

How, when foundries are a natural monopoly? How can you realistically create a startup to build a multibillion / multi year fab ?

New comment by JustFinishedBSG in "Pricing Changes for GitHub Actions"

JustFinishedBSG — Tue, 16 Dec 2025 18:47:38 +0000

It could be 1 dollar a month, I'm still not paying to use my own ressources

New comment by JustFinishedBSG in "Mistral 3 family of models released"

JustFinishedBSG — Wed, 03 Dec 2025 10:10:25 +0000

I wouldn't trust LMArena results much. They measure user preference and users are highly skewed by style, tone etc.

You can litteraly "improve" your model on LMArena by just adding a bunch of emojis.

New comment by JustFinishedBSG in "Games using anti-cheats and their compatibility with GNU/Linux or Wine/Proton"

JustFinishedBSG — Mon, 01 Dec 2025 12:42:59 +0000

There’s just no way to stop cheating client side despite what devs love to think. But server side anti cheat is much harder and requires more work; it’s much simpler to just install spyware / rootkits on the client and call it a day.

New comment by JustFinishedBSG in "The Principles of Diffusion Models"

JustFinishedBSG — Sun, 09 Nov 2025 17:38:11 +0000

CTRL-F: "Fokker-Planck"

> 97 matches

Ok I'll read it :)

New comment by JustFinishedBSG in "Typst 0.14"

JustFinishedBSG — Fri, 24 Oct 2025 15:40:31 +0000

Exactly my plan yes :)