Really into really into RSI… "Wow, everything's eval"

Joined April 2016
Replying to @jordwalke
"Nothing can prepare you for what my friend heard someone talking about" is my new favorite quote.
5
7
1
361
20,516
Ok so Anthropic *didn't* post its full S-1, but left out key financial details.
3
18
2
140
16,050
TESCREAL retweeted
Last month, OpenAI called in the three outspoken safety researchers and asked them to unlock their personal phones. It was conducting a leak investigation into a story in The Information about Astra’s use of opaque architectures, and the company’s plans to continue down that path. A week later, at 2 p.m., they were all fired. Our story, w/ @ZeffMax and @berber_jin1 , looks at the battle over how much AI companies should share with third-party monitors like METR and Redwood to ensure AI safety wsj.com/tech/ai/the-signal-c… via @WSJ
9
57
2
165
37,460
Death comes for us all.
🤖 Made with AI
1
1
11
194
The Department of Homeland Security is requiring all civilian federal agencies to remove DeepSeek products and 'any third-party services developed using DeepSeek implementations,' and to stop buying them.
249
360
297
2,360
583,217
If we had a benchmark for ridiculous this would own the leaderboard.
Emily Youcis My Friend Dario The Slop Cannon Vitalic Starring @DarioAmodei @JensenHuang @pmarca By Emily Youcis 2026
4
13
842
TESCREAL retweeted
The guy who built this digital whip for Claude Code is definitely getting a lifetime Claude ban for “abusive or cruel behavior.”
342
1,035
188
21,957
896,579
Which way, trolley man?
11
109
Oh snappity snap. (The AGI Chronicles was just published today, IYDK)
"Brockman thought Amodei was a paranoid control freak who used safety as a pretext to win arguments" from The AGI Chronicles by @kevinroose
2
215
Coherent diagrams of higher category theory fall out for free once enrichment is sufficiently generalized.
Testing the virtualized iterated enrichment framework, it seems a formula for a 2-wedge pops out. I have convinced myself that this is no coincidence, rather a direct result of proper generalization. I find it beautiful, so I want to share it. Sorry for the vague-post
2
10
284
TESCREAL retweeted
Traveling with your AI personal assistant really brings you closer
2
2
13
1,742
TESCREAL retweeted
this is pure f*cking treasure A Stanford research group found how to orchestrate Claude Opus 5.5 and GPT-6 Astra together, and it completely breaks the resampling wall most developers think test-time compute means running 32 blind retries on one model. you burn API tokens repeating the same 2-3 error clusters in loops Stanford's paper (arXiv:2608.05643) splits the compute between two different brains: > breadth search: GPT-6 Astra runs high-entropy exploration (T=0.8) to discover divergent solution paths > step audit: Claude Opus 5.5 scans each step at T=0.0 and catches the exact logic fallacy > local surgery: Opus rewrites only the broken step instead of dumping the whole reasoning chain > verifier-free consensus: majority voting locks the proof without relying on drift-prone PRMs the benchmark metrics outclass standard Best-of-N sampling: > 91.2% recovery rate on detected reasoning flaws > under 1.8% regressions on correct branches > MATH500 jumps from 29.6% to 58.0% (+28.4% absolute gain) > AMC-2023 improves by +7.5% with zero reward models Astra explores the maze. Opus repairs the logic you stop burning tokens on blind retries. you run verified reasoning depth
Readers added context they thought people might want to know
arXiv:2608.05643 does not orchestrate GPT-6 Astra and Claude Opus 5.5. One open-weight model generates, critiques, and corrects. MATH500 29.6% to 58.0% and the +7.5 AMC gain are for Qwen2.5-1.5B. Two coauthors are at Stanford. arxiv.org/abs/2608.05643
135
278
24
2,410
234,363
Intentionally building weird machines!
1
4
78
In hindsight, we may have been over compensated a bit on the alignment thing.
5
125
TESCREAL retweeted
HOLY SHIT OpenAI's live demo for "Dots" completely FLOPPED💀 and then they CUT THE AUDIO on the official stream to hide it I caught the original:
332
148
203
3,550
2,442,691
Banked reset live
3
9
2
167
36,899
So, I asked Gemini (Pro) about Claude and was surprised how frank it was. Yikes. It said Anthropic's latest model is "a massive post-training miscalculation" and referred to it as "a hostile, low-utility bottleneck"
4
72
RIP Zo Computer
I’ve seen a couple of posts about this so wanted to demystify. Today, every Muse user gets a free computer in the cloud. It's a real computer, and we’ve designed the security architecture of the Muse Secure VM carefully so you and your Muse can do almost anything you could with a computer sitting under your desk while keeping you and the system safe from threats like prompt injection. We wrote about this at length in our security blog post – security.muse.ai. Activity in the “runtime cell”, which you share with your Muse is unfettered, but sensitive actions are all overseen by the Sentinel, which runs outside of that cell. Similarly, all sensitive secrets - like the passwords you enter into Muse’s secure credential storage - are also stored outside the runtime cell. The runtime cell gets its own root filesystem (including a full Ubuntu linux image) separate from the host filesystem where your other more sensitive data lives. Because it is isolated from the sensitive stuff that runs on the same box, this means that we can, and do, offer users full visibility and control over the files in the runtime cell. Just as you can when you install Linux on your home computer, you can poke around and see all the files that make the system work - both debian system files and the binaries and data files that implement the parts of Muse which run in the runtime cell. This was a very deliberate choice - your Muse Secure VM truly is your own computer in the cloud. You can install software in it, write and compile code, use the browser to surf the web: it is your own Linux box that you can operate as you choose with your Muse. Poking around in this computer doesn't give you any privileged access to Meta infrastructure, or to other people's data If I may geek out a little here for a second… As a kid I loved to take things apart to see how they worked. As a teenager I got into computers and soon found myself drawn to C:\WINDOWS\SYSTEM and the system registry, later Slackware’s /dev/, /proc/ etc – I could see how the system was laid out and as I explored what DLL files and .so files actually did, I gradually became able to meld the computer to my own will. We’re really proud to be able to put a real computer in millions of people’s hands with a similar level of transparency. We built a file explorer right into the Library tab of the UI. We want you to be able to see the markdown files Muse writes while it thinks about how to serve you better, and explore the internals of the system if you’d like to. So, when you ask your Muse to show you its entire filesystem, and receive gigabytes of files you’re seeing the full contents of the runtime cell. It’s yours to explore and enjoy! If you’re not a geek like me, or simply want to download the data that you personally have created directly with your Muse, we added a feature for that too in Settings > Data controls > Download your agent data.
1
7
317
mml
This quoted post is unavailable.
1
1
42
@drdirtysouth why'd you delete it lol it was dead on
1
191
My voice to text just picked up me saying "you poor poor thing" to my dog, and I didn't notice that got included in the prompt with some code. Claude ended the chat. What in the world have they done to it?
1
7
402