kennywinker

8 ms·

91.
by kennywinker 1mo ago
https://artificialanalysis.ai/?models=qwen3-8-27b%2Cclaude-o... Index methodologies here: https://artificialanalysis.ai/evaluations/artificial-analysi... Also see some specific benchmarks here: https:
92.
by kennywinker 1mo ago
Jesus. Seems like your town should house that man.
93.
by kennywinker 1mo ago
It literally is usable now. A 5060 for $800 can run qwen3.8-27b 4bit at >40t/s, and the model beats opus 4.6 (max).
94.
by kennywinker 1mo ago
Claude, read Love in the Time of Cholera for me and summarize the information contained. You are an expert book reader and understander. Make no mistakes.
95.
by kennywinker 1mo ago
Another one to try: https://huggingface.co/Jackrong/Qwopus3.8-27B-Flash-GGUF Runs the 3bit model faster than the 2bit one runs on my old-ass card. Can’t vouch for its intelligence yet, but i suspect whatever loss in sm
96.
by kennywinker 1mo ago
Fair enough mental model, except my GPU can’t load the 8bit version and paging from disk makes it way more than 1/2 speed.
97.
by kennywinker 1mo ago
However strongly genetic marks predict criminality, zip code does it better. Any solution to crime that starts with eliminating types of people instead of poverty is just a form of fascism.
98.
by kennywinker 1mo ago
I think the author's point is that open weight models aren't going to be locked down like that. And even if they are locked down, it's hours between a model being released on huggingface and an "abliterated" variant
99.
by kennywinker 1mo ago
I mean, github pages using the github editor to edit docs pretty much fits that bill
100.
by kennywinker 1mo ago
I'm confused why the worry about LLMs that will answer "how do I build a pipe bomb". That information is easily available other places. The anarchist cookbook has been around and available for 55 years, and yet pipe bombs are
101.
by kennywinker 1mo ago
I can imagine one’s hat changing color when faced with a pot that big…
102.
by kennywinker 1mo ago
That should keep openai liquid for an extra 45min or so.
103.
by kennywinker 1mo ago
I didn't blame the administration at all. I don't even think this is specifically about political speech. You're ascribing a lot of beliefs to me that I haven't expressed, and assuming we have a similar experience of soc
104.
by kennywinker 1mo ago
"pro-hamas activists" is approximately nobody.
105.
by kennywinker 1mo ago
So free speech absolutists don't care about free speech, just free speech for americans on american soil? Not very absolutist then...
106.
by kennywinker 1mo ago
Ideologically vetting can mean a lot of things. Like, a sign on the door that says "NO TERFS" is a form of ideological vetting. By this logic any right wing message board would be designated terrorists the moment one person who ha
107.
by kennywinker 1mo ago
I will believe that's the goal when I see the tax going directly to aid in disposal.
108.
by kennywinker 1mo ago
"it" = people being banned for innocuous things. Just because you don't always get banned for political speech, doesn't mean random political speech bans don't have a chilling effect.
109.
by kennywinker 1mo ago
Before you go deep on OpenSCAD, consider JSCAD: https://jscad.app/ Roughly similar capabilities, but you get to use javascript or typescript instead of the linguistic abomination that is OpenSCAD.
110.
by kennywinker 1mo ago
I like that, and there are definitely situations where that describes what’s happening better. But I feel like adding intent changes it a bit. Some of the worst things exist not because of intent but because of indifference. E.g. nobody wa
111.
by kennywinker 1mo ago
I don’t. But given it didn’t change the last n times it happened, i’m doubtful. Whats makes you think it is being changed?
112.
by kennywinker 1mo ago
The purpose of a system is what it does. Incidents like this are common enough, that this is what the system does. If it’s AI, or rogue mods, or zuck himself, it doesn’t matter because this is what the system is doing - and if this wasn’t a
113.
by kennywinker 1mo ago
Right now? Nobody is using the hypothetical un-built datacenters that threaten to tax our planet. All the current individual LLM demand is being met by what exists already, yet we're seeing unprecedented amounts of money being spent at
114.
by kennywinker 1mo ago
Actual emissions reductions is different from just shell-gaming emissions from the US column to the China column. Greenwashing is real, but that doesn’t mean reducing emissions isn’t a real thing we can do.
115.
by kennywinker 1mo ago
Great point. For example, we should encourage collaboration between people building hyperscale data centers, and people not building hyperscale data centers to find some middle ground that is good for the planet. Collaboration implies we ar
116.
by kennywinker 2mo ago
Yes i think I acknowledged that, but is that useful for making a tool that can be trusted to safely run shell commands when asked arbitrary questions? No. It’s not.
117.
by kennywinker 2mo ago
> My sense following the news was that it was less arbitrary than it seemed at first What news have you seen that made it seem less like a retaliation?
118.
by kennywinker 2mo ago
Was that what that was about? Not punishing anthropic for denying them their killbots? Because it sure seemed like it was about punishing an entity that denied them something.
119.
by kennywinker 2mo ago
https://github.com/ThorOdinson246/whatisit-nl2sh uses a finetune of Qwen2.5-Coder-1.5B-Instruct. It works pretty well, tho it will misunderstand things from time to time
120.
by kennywinker 2mo ago
That’s true, but not true-true. Sure, every time you prompt “what is the weather in kansas” you’ll get the same output, but if you prompt “what is the weather in kansas right now” you’ll get a different output, and then “what is the weather

More ›