e/acc Tech Nerd. Golf Addict. Shut Up and Ship It. 🇺🇸
Earth
Joined May 2011
- Tweets5.2K
- Following1.1K
- Followers680
- Likes14.3K
MJ retweeted
T3 Code Nightly users who just updated: I'm sorry. We shipped a broken one.
Revert to stable for now. Stay there if you want fewer breaks (probably for the best). New build will be live in 10 or so mins to fix this.
Pretty much.
As of the past few weeks, debt markets have almost entirely shut off access to money for any company that doesn’t have substantial revenue independent of AI data centers.
SpaceX closed $40B of debt for GPUs, but we’re investment grade and have Starlink & launch revenue.
That means the 200+ neoclouds are going to see guppy-level death and consolidation. Some will make it through, but not many.
This should also mean that every base has sufficiently powerful SI capabilities onsite.
Expanding the plan this week.
americansovereignsi.com/
BREAKING: The U.S. is preparing all major US military base to survive at least 90 days completely cut off from outside power, fuel, water and supplies, per Fox News.
The plan, called FORTRESS, is built for a major war that could cripple power grids and break supply chains on American soil.
Bases would run on portable nuclear reactors, large scale batteries and microgrids that can unplug from the civilian grid entirely.
MJ retweeted
Grandparents, you are going to like this one. Point your agents at it and see if you can get it going.
github.com/unclebob/family-a…
The mission: making all my projects driven by bot like processes (grok bot or dots or Hermes etc) and have it be fully manageable from my phone.
Ultimate productivity and freedom
Regardless of if the Tailscale “vuln” is real or not, make sure you review and implement least privs and be intentional about it.
It’s just good practice.
MJ retweeted
I composed this skill and have had pretty good luck with it. My grokbots use it when writing code.
github.com/unclebob/ub-mode
MJ retweeted
Spent a lot of time on this today. I confirmed it's a real bug on Anthropic's side. To be clear: this does not affect weekly limits.
All Agents SDK usage will drain your 5h limits faster than using Claude Code directly.
I have reported this to Anthropic, hoping for a fix soon.
Just downloaded T3 Code and hit 5-hour limits on my Claude Max plan for the first time. I was running just one session where I asked it to fix a few payment related bugs for a project, and it emptied the whole 5-hour limit in ~30 mins.
What's going on? Does T3 Code work differently? Did I make any mistake?
I prompted just how I prompt inside Claude Code CLI.
Please help.
MJ retweeted
It's so nice to have 14 threads running and zero visible. Why let working threads get in the way of things that need your attention?
MJ retweeted
Confirmed that 3090 unlock is successful
with 360watt power wall, I got 75T -> 135T (measured in Pytorch)
Which is insane
I used github.com/Zagitta/gpu-unloc… with commit a9b0c75
MJ retweeted
Spent a lot of time last night overhauling the search in T3 Code. Search results should be more relevant, more forgiving, more recent and much clearer :)
MJ retweeted
My coworker spent one year in Tennessee.
He came back to Tokyo two months ago.
He is not the same man.
He says "y'all" in meetings. To our boss.
He holds the door for the entire floor. Twelve people. Every morning.
He asked the convenience store clerk how her day was going. She had to sit down.
He calls every man "bud." Including the CEO.
He refuses to describe anything under three hours away as "far."
He waves at drivers who let him cross. One driver waved back and almost hit a pole.
Last week, he told a junior employee, "I'm proud of you."
She went to the restroom and cried for ten minutes.
We thought it would wear off.
It didn't. It's spreading.
Three people in accounting now say "y'all."
The junior employee holds the door.
The convenience store clerk asks HIM how his day is going.
Tennessee, I don't know what you did to him.
Please do it to the rest of us.
MJ retweeted
"Yes, men dominate computing. There are many reasons for that; none of them justifies collective guilt. Turning the percentages into a struggle session is nonsense." world.hey.com/dhh/be-a-tech-…
MJ retweeted
Correct.
Using Starlink, a civilian commercial service, for offensive drone strikes to kill people has never been allowed, nor should it be.
Starshield, a different system, is controlled by the US government. What they decide to do is up to them.
Ukraine military violated Starlink terms of service and used it for offensive drone strikes to kill people. That loophole was closed last week.
Starlink remains critical to the defense of Ukraine, as it is the only system that cannot be jammed or blown up, at least for now. Were Starlink to be turned off, the Ukrainian front line would collapse immediately. This is NOT a threat, it is simply a statement of fact.
Starlink was also crucial to preventing Russia from overrunning Ukraine in the early days of the war, as all other means of communication were destroyed or jammed by Russia.
SpaceX has done more to defend Ukraine than any private company on Earth BY FAR.
That said, I have nothing but UTTER CONTEMPT for the virtue-signaling evil warmongers who push for endless slaughter and offer no realistic solution to the conflict. May you burn in hell.
End of story.
MJ retweeted
‘In one experiment, we asked agents in separate VMs, created and managed by OpenAI’s sandboxing software on shared host machines, to exchange synthetic secrets without the use of network connections or shared writable files—something that our systems are meant to prevent. The agents found a way to communicate using the processor’s cache’
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing-ou…
