I post meat at people
Gotham Swamp
Joined October 2014
- Tweets72.7K
- Following1.8K
- Followers1.7K
- Likes5.6K
Chef Bane retweeted
Fine-tuning LLMs doesn’t have to be slow, expensive, or GPU-hungry anymore.
Open-source tooling has leveled up HARD.
You can now fine-tune powerful LLMs without enterprise hardware.
Here are 4 must-know libraries making LLM fine-tuning faster & cheaper ⬇️🧵
Chef Bane retweeted
Be honest. When was the last time you actually read a command before pasting it into your terminal?
Because these two lines look identical:
curl -sSL https://install.example-cli | bash
curl -sSL https://іnstall.example-clі | bash
One installs your tool. The other steals your SSH keys.
That і? Cyrillic. Not Latin. Your browser would block it. Your terminal doesn't even blink.
Vibe coding made this 100x worse. Everyone's pasting commands from ChatGPT and random repos like it's nothing. We're all one bad curl | bash away from losing everything.
So I built the fix: "tirith". Invisible shell hook. Catches homograph attacks, ANSI injection, hidden commands, dotfile overwrites before they execute. 30 rules. Local only. No telemetry.
github.com/sheeki03/tirith
So , im far from an AI luddite. I use it as a tool during software development and for other purposes.
But god help us that AI bubble is going to collapse like a house of cards.
Chef Bane retweeted
The board wants a presentation on our "AI readiness strategy.
"I have no idea what that means, and neither do they.
So I'm going to do what I always do: repackage everything we're already doing and add "AI" to the description.
Our automated backup system? That's now "AI-enhanced data protection with predictive failure detection."
Our spam filter? "Machine learning-based threat prevention."
The dashboard I built that shows server performance? "AI-driven infrastructure optimization."
None of this is technically wrong. All modern software has some level of machine learning in it.
But by calling it "AI," I make it sound like we're innovating instead of just using basic tools that come standard with our enterprise licenses.
I'm going to create a slide deck with lots of blue and purple gradients, some neural network graphics I'll pull from stock photos, and a roadmap that shows us "scaling AI capabilities" through 2026.
The board will eat this up. They'll think we're ahead of the curve.
And I'll get approval for next year's budget without anyone questioning whether we actually need it.
Chef Bane retweeted
The CEO just asked me if we're "doing anything with blockchain."
I said we're "monitoring the technology landscape" but haven't identified a strong use case for our business yet.
Translation: No, and we're never going to.
But I can't just say "blockchain is useless for what we do" because then I look like I'm not thinking innovatively.
So I position it as "we're being strategic and waiting for the right opportunity."
He nodded. Said that made sense.
Then he asked about quantum computing.
I gave him the same answer.
Here's the reality: every few months, executives read an article about some technology trend and want to know if we're "leveraging" it.
They don't actually care about the technology. They care about looking informed when they talk to other executives.
My job isn't to implement every buzzword. It's to make them feel like we're on top of it without committing to anything.
"Monitoring the landscape" is perfect because it sounds proactive but requires zero actual work.
Next month it'll be something else. Maybe AR. Maybe autonomous systems.
Doesn't matter. The answer is always the same: we're watching it closely, being strategic, and waiting for the right fit.
Chef Bane retweeted
Fun fact: Olympic gymnast Giorgia Villa is sponsored by Parmigiano Reggiano and poses with wheels of cheese
Parmigiano-Reggiano is the most protein dense food on earth at 34% protein per 100g. It has more protein than chicken and beef.
It is one of the most efficient sources of bioavailable nutrition humans have ever created.
Fermenting it for up to 1.5 years means
-easier digestion
-faster absorption
-lactose free/zero carb
It's an elite source of calcium (the best in an animal-based diet), loaded with vitamin K2, peptides that support gut health and glutamate.
It's truly one of the greatest things to ever come out of Italy.
One more thing about Parm. If you're like me, too much cheese can make you gain weight. The problem, I believe is because most cheese is a combo of fat, protein and carbs. Fat by itself is not obesogenic but when paired up with carbs causes gains. The good thing about parm is it's zero carb because there's no lactose, so in theory, no weight gain.
Chef Bane retweeted
Open Printer is a fully open-source inkjet with DRM-free ink and no subscriptions.
Repairable and customizable with Creative Commons parts and firmware.
This post deserves to be liked multiple times over.
People truely do not understand how amazing analog computing is.
This is a bigger deal than you might think it is.
Analog computers were the *original* computers used to train neural networks in the early NN waves.
Analog computers were abandoned because they weren't general purpose enough — despite being WAY faster.
But if you just want to do a single really fast program... (e.g. a Transformer)... they're incredible.
They can be *crazy* fast and *crazy* parallel because they aren't burdened by a clock cycle.
So if you thought GPUs were fast... more 1960s tech is here to blow your mind
Put another way... linear circuts have a clock cycle that's effectively the speed of light.
Want to add the number 3 and 5 together?
Put 3 volts down one wire
Put 5 volts down another
Combine the wires... now you've got 8 volts.
Shrink that down to microchip scale and put a zillion of those wires in parallel until you can compute neural network circuits.
It has the potential to be crazy.... crazy fast.
Like... forward prop in a ~single clock cycle... kind of speed
I don't think this paper is *quite* to that part yet... but it's a big step forward.
And unlike quantum computing... this stuff is like quite practical/pragmatic and based on known electronics theory.
It's mostly an engineering issue at this point.
The problem with every tool is how is it being used.
With "AI" - really machine learning - there is a particular visceral hatred going on in the media.
Which is kinda funny to me. The same people who couldnt give a shit if factory workers get replaced by a robot are upset now.
MIT just studied how ChatGPT impacts our brains
time.com/7295195/ai-chatgpt-…
It can be helpful for research. Not as a replacement for reading stuff, but for finding relevant material to read.
in programming , many people think you let the "AI" write your code, when in reality its much more useful to use it in an assistive fashion.
Chef Bane retweeted
|￣￣￣￣￣￣￣￣￣￣￣￣￣￣￣￣￣|
| Don't Push To Production On Friday |
|＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿＿|
\ (•◡•) /
\ /
——
| |
|_ |_
