Mal

I build things. Good design excites me. (he/him) 🎶🌊🌴🍺

[Mal @mtripn]

"Shake what your mama gave ya"

*shakes existence*

[auror ✞ @hoeror]

Dix-huit mille quatre cent cinquante-sept. Et on ne parle QUE des ENFANTS

[ALUTHEDON @Mbakaza4L]

Fact: He traveled there for his cooking show & when they tried to block this one episode, he said he would pull his entire show from the network & this was the first time the world ever saw behind the walls in Gaza 2013

[Ihab Hassan @IhabHassane]

Sharing a photo from two years ago — when our people in Gaza were struggling to find a single piece of food during famine — to dehumanize them is what’s truly sick here.

You seem to care more about a dolphin than about starving children.

[pokey pup @Whatapityonyou]

Bombing children wasn’t horrendous enough apparently

[George Carlin's Ghost @slayguevara]

Your ancestors literally couldn't dream of this level of civilisation you troglodyte

[Mohamad Safa @mhdksafa]

Imagine if Iran bombed and destroyed the Golden Gate Bridge in California, what would you call it? TERRORISTS

The U.S bombed and destroyed the highest bridge in Iran, the B1 bridge, why do you call it “PEACE”?

Bombing a bridge is a war crime under international humanitarian law

[Mel @Villgecrazylady]

They are everything Hollywood told you the Nazis were

[Ali Abunimah @AliAbunimah]

Hats off to whoever authored this tweet

[Martha Lincoln @heavyredaction]

This staggering act of ecocide is not even receiving mainstream coverage

[Karim @Idlibie]

Nice AI slop.

Now here’s a real picture of the 165 graves of Iranian girls who will never reach womanhood because you murdered them all only days ago.

[Vindhya Buthpitiya @vindib_]

our countries are not your retirement colonies.

[BURKOV @burkov]

LLMs process text from left to right — each token can only look back at what came before it, never forward. This means that when you write a long prompt with context at the beginning and a question at the end, the model answers the question having "seen" the context, but the context tokens were generated without any awareness of what question was coming. This asymmetry is a basic structural property of how these models work.

The paper asks what happens if you just send the prompt twice in a row, so that every part of the input gets a second pass where it can attend to every other part. The answer is that accuracy goes up across seven different benchmarks and seven different models (from the Gemini, ChatGPT, Claude, and DeepSeek series of LLMs), with no increase in the length of the model's output and no meaningful increase in response time — because processing the input is done in parallel by the hardware anyway.

There are no new losses to compute, no finetuning, no clever prompt engineering beyond the repetition itself.

The gap between this technique and doing nothing is sometimes small, sometimes large (one model went from 21% to 97% on a task involving finding a name in a list). If you are thinking about how to get better results from these models without paying for longer outputs or slower responses, that's a fairly concrete and low-effort finding.

Read with AI tutor: https://t.co/MipHHO6rjX

Get the PDF: https://t.co/XQrqiaGwIO

[Air Katakana @airkatakana]

@burkov the research papers of 2025 are the twitter posts of 2023

[Avery Edison @aedison]

sigh

[安天美 @tomiadesina_]

why I get aggravated when people cut trees because they need to “develop” an area.
just say you lack creativity and vision 😔.

[Judicaelle Irakoze @Judicaelle_]

Globalization really makes citizens of first world countries go occupy lands and communities in the third world, turn them into unlivable places for the locals, then leave for those same locals to fend for themselves through hard times.

​the tweet below is a good example. Ghana is also another example.

[Mal @mtripn]

@tylerpalmer @paulg You sweet summer child

[Mal @mtripn]

@IanFortier @ste_bau Maybe Ian’s point isn’t just about this post, but about Steve choosing to break his “not political” stance only once someone known for fueling hate against marginalized groups was gone.

[patagucci perf papi @kenwheeler]

@sama “you stole the entire internet under the guise of an open foundation and you’re selling it back to us if we wanna keep our jobs, fuck you”

[Mal @mtripn]

@teej_dv @excalidraw This will imply the existence of an excalidrawest