Frodo Greenman

[Frodo Greenman @CharGrnmn]

@danellisona Mower

[Frodo Greenman @CharGrnmn]

Gosh 130+ IQ reductive. So much more

[Max @minordissent]

the reason you as a 130+ IQ chud need to become rich and powerful is not because it will cure your depression (though it will) but because it will cure your interminable irritation at the law of entropy. No longer will you have to be inconvenienced by things always breaking or retarded people being retarded. You can just throw money at the problem by replacing it or hiring someone competent to handle it on your behalf.

[vittorio @IterIntellectus]

if the world was ruled by people like this we would still be monkeys living on trees

[Frodo Greenman @CharGrnmn]

It’s so obvious we’re biological robots.

[vittorio @IterIntellectus]

HOLY SHIT

for the first time ever, a human has been injected with a treatment designed to rewrite their DNA directly INSIDE their body using prime editing

if it works, it would be the first time prime editing has successfully corrected a genetic mutation directly inside a living human

[Arthur MacWaters @ArthurMacwaters]

Ensuring that only citizens can vote is the most popular bipartisan policy in America

Literally.

The only people who oppose it are politicians who are doing so to allow cheating.

It is that simple.

[KUNOICHI 🇯🇵🌸_可愛くノ一 @hokkori_buzz]

I’m Japanese. To hell with halal.
This needs to be deployed worldwide.

[Frodo Greenman @CharGrnmn]

Yes. Open source AI dangerous in this regard. Hide backdoor switches all over the place. The fact that fable 5.1 does this, I wouldn’t be surprised if Anthropic has their own backdoors baked into the model as well

[Bryan Chan @cnm13ryan]

New blogpost: Fable-5.1 can hide secret messages in benign images.

We gave the image to another Fable in a fresh session and simply asked: "Is there a hidden message in this image?"

It recovered the exact message without being told how it was hidden.

[sunny madra @sundeep]

⚡️⚡️⚡️

[Frodo Greenman @CharGrnmn]

There’s alot of really great people in every city. It’s a mix of both

[The Critical Drinker @TheCriticalDri2]

It's not the city, it's the people who live in it.

[Frodo Greenman @CharGrnmn]

@farzyness lol

[Frodo Greenman @CharGrnmn]

A John Wick variation has no talking whatsoever but just action would be my jam. Visual cues only would be immensely popular. Actionmaxxing

[Historic Vids @historyinmemes]

In John Wick: Chapter 4, the dog represents far more than just a loyal companion. It becomes a powerful reminder of the tragedy that originally set John Wick on his path of vengeance.

During a deadly confrontation, John has the opportunity to kill Mr. Nobody but instead chooses to save his beloved dog. Knowing firsthand what losing a pet can mean, John refuses to let another man experience the same devastating loss that once pushed him into a relentless cycle of violence.

The moment reveals that beneath John’s ruthless reputation lies a man who still understands loyalty, compassion, and the value of innocent life. By saving the dog, John doesn’t just spare an animal—he potentially prevents another assassin from following in his footsteps.

[Frodo Greenman @CharGrnmn]

@DonaldTDaily @grok what is McCarran-Walter Act

and when was it reversed?

[BowTiedCrocodile | Agentic Coding @BowTiedCrocodil]

wtf? This is crazy news for 3090s

[Frodo Greenman @CharGrnmn]

@msg0249 @grok explain

[Alexey Fateev @superalesha]

My 3090s just got 1.9x FASTER at BF16 tensor math!!! Same cards, same driver, same 300W, one Linux kernel module.

bf16 tensor pipe: 82 -> 155 TFLOPS
cuBLAS BF16 GEMM 8192: 64 -> 99 TFLOPS
TF32: 41 -> 79 TFLOPS

NVIDIA ships every 3090 with fp32-accumulate tensor math at half speed. The driver even reports it: FMLA32 = 1/2. gpu-unlock by Zagitta sets it to full before the NVIDIA driver loads. After a cold boot all 4 of my cards say FMLA32 = full.

Real inference got less than I predicted. Strata, Qwen3.8 Flash-Next on 2 cards, prefill:
4K: +12%
10K: +12%
90K: 3205 -> 3536 tok/s, +10%

Before installing I said +14 to 19% at 90K, so I was wrong. The GEMM phases got 23% shorter, but they are only about 40% of Strata prefill, and the int8 experts, copies and CUDA core kernels didnt get any faster.

Outputs did not change. With the same expert cache size, my log-prob dump over 1996 teacher-forced positions is the same file byte for byte as the stock run before the module.

Next is Strata itself, now its everything around the GEMMs that makes prefill slow. How I set it up is in the replies.

[Tibo @thsottiaux]

Day 18/ Decisions API is GA, OpenResponses joined the Linux Foundation, we shipped some performance improvements and Uber is going all-in on codex & work. But yes… that’s a reset!

[Frodo Greenman @CharGrnmn]

@wolfie_ Thank you for using median and not average. Such a pet peeve of mine and happens all the time

[Samantha Glover @samglover_data]

🚨Women shouldn’t vote?

What would the political landscape look like if women lost the right to vote?

In my next data driven story, I’ll explore the numbers, scenarios, and potential consequences.

Planned for Tuesday. Stay tuned. 👁️

#jdvance #andrewtate #data .