HNHacker News
TopNewBestAskShowJobs

gdevenyi

347 karma · joined August 29, 2018

submissionscomments
gdevenyi··on Residential proxies detected on White House network
It would be really great to have some info about how to detect such things locally.
gdevenyi··on WTF Ubuntu?
I used to be a contributor to distros, but I found that Ubuntu simply has no method by which I can fix things and contribute them back.
gdevenyi··on WorldTUI: I trained a 360M model to pretend to be a terminal
Perfect honeypot
gdevenyi··on Apple HomeView
> an old ipad

You've just realized they're going to obsolete all those products.

gdevenyi··on Crest – an Arc-inspired open-source browser for Mac
This is the result of so many very vibe developers on Mac.
gdevenyi··on WSL3 Performance is about 5-60% faster than WSL2 depending on the workload
Could get all the performance back if you just ran Linux natively.
gdevenyi··on Run Qwen 3.8 Flash Next (125B) on consumer hardware (RTX 4090) at 100T/s
I had this working with the FreeToken inference engine a month ago when they launched.

https://github.com/FlashML-org/FreeToken

gdevenyi··on Linux support is coming to Snapdragon X2 Series
This sounds like a terrible situation, but I wonder if a weekend of Claude iterating could improve on any of these problems. They seem to have been pretty successful on the Mac M series.
gdevenyi··on AI Protest in Montreal
You got it
gdevenyi··on AI Protest in Montreal
> They argue that LLMs are stealing our culture to feed it back to us.

How is that any different from what media already did.

Most of Disney's original material was European cultural stories.

gdevenyi··on Ubuntu 26.10 completes transition to Rust-based coreutils
Reporting bugs _at all_ in Ubuntu has never worked for me.
gdevenyi··on Why is Google still serving dodgy ads?
People still see ads?
gdevenyi··on Fable 5.1 Solves the Cyphral Distich, a 370-year-old cipher
Anyone pointed it at the Voynich manuscript?
gdevenyi··on Homebrew 7.0.0
Does it still need root on Linux? That makes it an instant no go.
gdevenyi··on Smolts: A pedagogical IDE for a teaching language
Who remembers turning and object oriented turing from the university of Waterloo?
gdevenyi··on A mysterious kidney disease has arrived in Texas
https://archive.ph/h66Jg
gdevenyi··on 4.5B Posts Scraped from TikTok
Like gazing into hell
gdevenyi··on Where did all the public bathrooms go?
https://www.jstor.org/stable/41212064
gdevenyi··on Seeing beyond BMI: Estimating cardiometabolic risk with smartphone imagery
Google already has a product called PhotoScan.

Come on guys!

gdevenyi··on Firefox is now the last major browser that still supports uBlock Origin
Brave has the same style of engine built into the browser.
gdevenyi··on Taxi drivers rarely die of Alzheimer's
People who are likely to get Alzheimer's don't get jobs as taxi drivers.
gdevenyi··on GTX 1080s: Testing a Legend
My 1080ti is my local inference machine now. With the release of the Bonsai 27B models I can run a genuine dense model with usable context. GOAT card.
gdevenyi··on Amber the programming language compiled to Bash/Ksh/Zsh
> Typescript to bash

Literally the worst of both worlds.

gdevenyi··on Show HN: ZeroFS – A log-structured filesystem for S3
There is some kind of advertising push for this going on. I got Twitter ads for this repo.
gdevenyi··on Podman v6.0.0
Because they don't publish up to date packages for major distros.
gdevenyi··on Too many R packages: CRAN is inundated with submissions
As a sysadmin R is the absolute worst for installing and supporting packages.
gdevenyi··on Matrix Multiplications on GPUs Run Faster When Given “Predictable” Data (2024)
People have been noticing the effects of this in local LLM inference. Power limiting seems to improve overall performance!
gdevenyi··on Qwen3.6-35B-A3B speculative decoding is net-negative on RTX 3090
I'm very interested if more VRAM limited setups (8-16GB) show the same performance loss or if they react differently.
gdevenyi··on UK Biobank health data keeps ending up on GitHub
This.
gdevenyi··on Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model
Mine does https://github.com/gdevenyi/huggingface-estimate
Page 1 of 6Next →