frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Plasma Bigscreen – 10-foot interface for KDE plasma

https://plasma-bigscreen.org
74•PaulHoule•58m ago•19 comments

this css proves me human

https://will-keleher.com/posts/this-css-makes-me-human/
125•todsacerdoti•3h ago•45 comments

C# strings silently kill your SQL Server indexes in Dapper

https://consultwithgriff.com/dapper-nvarchar-implicit-conversion-performance-trap
40•PretzelFisch•2h ago•26 comments

Show HN: The Roman Industrial Revolution that could have been (Vol 2)

https://thelydianstone.com/volume-2
20•miki_tyler•1h ago•11 comments

Math Notepad

https://mathnotepad.com
51•the-mitr•4d ago•17 comments

The Shady World of IP Leasing

https://acid.vegas/blog/the-shady-world-of-ip-leasing/
46•alibarber•3h ago•22 comments

Hardening Firefox with Anthropic's Red Team

https://www.anthropic.com/news/mozilla-firefox-security
464•todsacerdoti•13h ago•133 comments

Show HN: Moongate – Ultima Online server emulator in .NET 10 with Lua scripting

https://github.com/moongate-community/moongatev2
219•squidleon•10h ago•129 comments

Open Camera is a FOSS camera app for Android

https://opencamera.org.uk/
206•tetris11•4d ago•101 comments

Apache Otava

https://otava.apache.org/
78•djoldman•5d ago•5 comments

Tech employment now significantly worse than the 2008 or 2020 recessions

https://twitter.com/JosephPolitano/status/2029916364664611242
693•enraged_camel•7h ago•471 comments

Launch HN: Palus Finance (YC W26): Better yields on idle cash for startups, SMBs

34•sam_palus•6h ago•54 comments

The worst acquisition in history, again

https://www.profgmedia.com/p/the-worst-acquisition-in-history
85•JumpCrisscross•3h ago•62 comments

Payphone Go

https://walzr.com/payphone-go/
305•walz•4d ago•67 comments

Galileo's handwritten notes found in ancient astronomy text

https://www.science.org/content/article/galileo-s-handwritten-notes-found-ancient-astronomy-text
14•tzury•1d ago•0 comments

Show HN: Kula – Lightweight, self-contained Linux server monitoring tool

https://github.com/c0m4r/kula
3•c0m4r•50m ago•0 comments

Ada 2022

https://www.adaic.org/ada-resources/standards/ada22/
102•tosh•4h ago•19 comments

CT Scans of Health Wearables

https://www.lumafield.com/scan-of-the-month/health-wearables
177•radeeyate•10h ago•39 comments

Triplet Superconductor

https://www.sciencedaily.com/releases/2026/02/260221000252.htm
47•jonbaer•4d ago•9 comments

Art Bits from HyperCard

https://archives.somnolescent.net/web/mari_v2/junk/hypercard/
50•TigerUniversity•3h ago•13 comments

Entomologists use a particle accelerator to image ants at scale

https://spectrum.ieee.org/3d-scanning-particle-accelerator-antscan
94•gmays•9h ago•16 comments

Multifactor (YC F25) Is Hiring an Engineering Lead

https://www.ycombinator.com/companies/multifactor/jobs/lcpd60A-engineering-lead
1•multifactor•7h ago

Astra: An open-source observatory control software

https://github.com/ppp-one/astra
77•pppone•8h ago•20 comments

LibreSprite – open-source pixel art editor

https://libresprite.github.io/
249•nicoloren•15h ago•82 comments

A tool that removes censorship from open-weight LLMs

https://github.com/elder-plinius/OBLITERATUS
109•mvdwoord•10h ago•44 comments

I'm 60 years old. Claude Code has ignited a passion again

19•shannoncc•52m ago•5 comments

The disappearing Form D (2018)

https://techcrunch.com/2018/11/07/the-disappearing-form-d/
8•eatonphil•2d ago•0 comments

Workers who love ‘synergizing paradigms’ might be bad at their jobs

https://news.cornell.edu/stories/2026/03/workers-who-love-synergizing-paradigms-might-be-bad-thei...
499•Anon84•11h ago•285 comments

Anthropic, please make a new Slack

https://www.fivetran.com/blog/anthropic-please-make-a-new-slack
178•georgewfraser•5h ago•156 comments

Analytic Fog Rendering with Volumetric Primitives (2025)

https://matejlou.blog/2025/02/11/analytic-fog-rendering-with-volumetric-primitives/
77•surprisetalk•1d ago•6 comments
Open in hackernews

A tool that removes censorship from open-weight LLMs

https://github.com/elder-plinius/OBLITERATUS
109•mvdwoord•10h ago

Comments

greenpizza13•6h ago
Never stopped to ask if they should...
Alifatisk•5h ago
This is for local models right? I can't use it on, say my glm-5 subscription connected to opencode?
HanClinto•4h ago
Correct, local models only.
ComputerGuru•5h ago
Reviews of the tool on twitter indicate that it completely nerfs the models in the process. It won't refuse, but it generates absolutely stupid responses instead.
Animats•5h ago
Link?

It's interesting that people are writing tools that go inside the weights and do things. We're getting past the black box era of LLMs.

That may or may not be a good thing.

noufalibrahim•5h ago
I believe that this is already done to several models. One that I've come across are the JOSIEfied models from Gökdeniz Gülmez. I downloaded one or two and tried them on a local ollama setup. It does generate potentially dangerous output. Turning on thinking for the QWEN series shows how it arrives at it's conclusions and it's quite disturbing.

However, after a few rounds of conversation, it gets into loops and just repeats things over and over again. The main JOSIE models worked the best of all and was still useful even after abliteration.

thegrim33•4h ago
Whether or not the linked tool uses a good approach, manipulating models like you mention is already fairly well established, see: https://huggingface.co/blog/mlabonne/abliteration .
kube-system•5h ago
I guess it's kind of like a lobotomy tool.
sheepscreek•4h ago
I guess it proves you cannot unlobotomize a hole in the head.
littlestymaar•5h ago
This is vibecoded garbage that the “author” probably didn't even test by themselves since making this yesterday, so it's not surprising that it's broken.

Also, as I said in a top level comment, what this project wants to achieve has been done for a while and it's called Heretic: https://github.com/p-e-w/heretic

(Not vibecode by a twitter influgrifter)

quotemstr•4h ago
We will eventually arrive at a new equilibrium involving everyone except the most stupid and credulous applying a lot more skepticism to public claims than we did before.

And yeah, doing stuff like deleting layers or nulling out whole expert heads has a certain ice pick through the eye socket quality.

That said, some kind of automated model brain surgery will likely be viable one day.

dinunnob•4h ago
Hate to have to be the one to stick up for pliny here, but hes concerned about forcing frontier labs to focus more on model guardrails - he demonstrates results that are crazy all the time

https://x.com/elder_plinius

IncreasePosts•4h ago
I didn't use this tool, but I did try out abliterated versions of Gemma and yes, it lost about 100% of it's ability to produce a useful response once I did it
electroglyph•1h ago
the default heretic with only 100 samples isn't very good, you really need your own, larger dataset to do a proper abliteration. the best abliteration roughly matches a very careful decensor SFT
halJordan•2h ago
Everyone says that abliteration destroys the model. That's the trope phrase everyone who doesn't know anything but wants to participate says. If someone says it to you, ignore them.
butILoveLife•1h ago
This is my experience with abliterated models.

I use Berkley Sterling from 2024 because I can trick it. No abliteration needed.

littlestymaar•5h ago
Don't use this 2 days old vibe coded bullshit please.

p-e-w's Heretic (https://news.ycombinator.com/item?id=45945587) is what you're looking for if you're looking for an automatic de-censoring solution.

a2128•5h ago

    You're not just using a tool — you're co-authoring the science.
This README is an absolute headache that is filled with AI writing, terminology that doesn't exist or is being used improperly, and unsound ideas. For example, it focuses a lot on doing "ablation studies", by which it means removing random layers of an already-trained model, to find the source of the refusals(?), which is an absolute fool's errand because such behavior is trained into the model as a whole and would not be found in any particular layer. I can only assume somebody vibe-coded this and spent way too much time being told "You're absolutely right!" bouncing back the worst ideas
creatonez•4h ago
> For example, it focuses a lot on doing "ablation studies", by which it means removing random layers of an already-trained model, to find the source of the refusals(?), which is an absolute fool's errand because such behavior is trained into the model as a whole and would not be found in any particular layer.

That doesn't mean there couldn't be a "concept neuron" that is doing the vast majority of heavy lifting for content refusal, though.

mapontosevenths•15m ago
Thats not what it means at all. It uses SVD[0] to map the subspace in which the refusal happens. Its all pretty standard stuff with some hype on top to make it an interesting read.

Its basically using a compression technique to figure out which logits are the relevant ones and then zeroing them.

[0] https://en.wikipedia.org/wiki/Singular_value_decomposition

dinunnob•4h ago
Hmm, pliny is amazing - if you kept up with him on social media you’d maybe like him https://x.com/elder_plinius
bigyabai•4h ago
If this qualifies as "amazing" in 2026 then Karpathy and Gerganov must be halfway to godhood by now.
dinunnob•3h ago
I dont think anyone is going to dispute this
bigyabai•3h ago
I just don't think many people will be "amazed" by their output, as you claim.
dinunnob•3h ago
I just said pliny was amazing, fwiw - i like that hes hacking on these and posts about it. I rushed to defend, i wish more people were taking old school anarchist cookbook approaches to these things
cess11•2h ago
Smoke banana peel?
Zetaphor•1h ago
I had such a godawful headache from that. Also tried the peanut shells, equally awful. I was a dumb teenager.
fragmede•1h ago
gasoline and styrofoam was fun tho
EGreg•3h ago
Amazing as in his stuff actually works?

I just hear him promoting OBLITERATUS all day long and trying to get models to say naughty things

dinunnob•3h ago
Yeah but i think the philosophy is to show how precarious the guardrails are
gavinray•3h ago
The parent comment makes no reference to or comment on the author of the README.

It just says "the README sucks." Which, I'm inclined to agree, it does.

LLM-generated text has no place in prose -- it yields a negative investment balance between the author and aggregate readers.

Aurornis•39m ago
I don't know. I scrolled through his recent Tweets and he's sharing things like this $900 snake oil device that "finds nearby microphones" and "sends out AI-generated cancellation signals" to make them unable to record your voice : https://x.com/aidaxbaradari/status/2028864606568067491

Try to think for a moment about how a device would "find nearby microphones" or how it would use an AI-generated signal to cancel out your voice at the microphone. This should be setting of BS alarms for anyone.

It seems the Twitter AI edgey poster guy is getting meta-trolled by another company selling fake AI devices

robertk•4h ago
You don't know what you are talking about. Obviously refusal circuitry does not live in one layer, but the repo is built on a paper with sound foundations from an Anthropic scholar working with a DeepMind interpretability mentor: https://scholar.google.com/citations?view_op=view_citation&h...
paradox460•2h ago
It's not just a headache, it's bad
Retr0id•1h ago
I don't know if this particular tool/approach is legit, but LLM ablation is definitely a thing: https://arxiv.org/abs/2512.13655
fragmede•1h ago
Alternately, it's intentional. It very effective filters out people with your mindset. You can decide if that's a good thing or not.
eli•1h ago
Why would a tool that works need to dissuade skeptics from trying it?
lazzlazzlazz•22m ago
Ironic to see this comment when Pliny, the author of this codebase, is one of the most sophisticated LLM jailbreakers/red-teamers today. So presumptive and arrogant!
measurablefunc•4h ago
This is another instance of avant-garde "art".
PeterStuer•4h ago
Already censored for sharing on FB Messenger?
ftkftk•3h ago
Didn't make it past the first paragraph of AI slop in the README. Have some respect for your readers and put actual information in it, ideally human generated. At least the first paragraph! Otherwise you may as well name it IGNOREME.
SilverElfin•2h ago
Does anyone offer a live (paid) LLM chatbot / video generation / etc that is completely uncensored? Like not requiring doing any work except just paying for it?
nomel•2h ago
Grok was one of the closest, with expected results: bad PR from the obvious use cases that come with little censorship.
mapontosevenths•6m ago
Nous Hermes was built from the ground up to be uncensored. No alliteration required.

Its not a frontier model but it will give you a feel for what its like.