🌯

[🌯 @invocationn]

@Cornelius_MacG @UnderdogCFB the defender knows that he cant stop them from getting the first down nonetheless he follows through with the tackle. i think that way he put his head down makes me thing he had time to pull up

[Paully B @boringbuildr]

Me at the red light while the homeless guy stares in my window

[ken @greeen2me]

accidentally hit my insanely burnt vape instead of the new one and got a mouth full of chatgpt dick

[⟢ @R6IVE5]

donald trump cutting down the cherry blossom trees gifted to us by japan from over a century ago for the sole purpose of building a golf course....

[Pokemon Gems @PokemonGems]

two babies meet

[xixi ✿ @KH4LDR0G0SWCW]

“I am heartbroken 💔 “

[¥ @p0xtahprv]

DO THE FUCKING DANCE NOW PUSSY MAGA FC DAANCE NOWWWWWWWWWWWWWW

[Tío Mendy @mendxx_23]

Es el mejor video de la historia

[Parlay Papi @KobesAndJordans]

https://t.co/7IkAmXHYQg

[UI/UX Savior @UiSavior]

🤔

[cleytxn do tuiter @cleytxnn]

URGENTE: Carlo Ancelotti cogita MATAR Neymar para conseguir "revivê-lo" sem lesões, com força total e possuindo energia infinita através do Jutsu Edo Tensei.

[🌯 @invocationn]

@Underdog 🫡

[kuz @kylekuzma]

Quantum compute.

[Mac🦬 @tha_buffalo]

“We’re going to build a data center in Cleveland”

Cleveland:

[WarrenBuffering @WarrenInTheBuff]

ERRRRRBODY WANNA BE A PROGRAMMER, NOBODY WANNA READ THE SHITTY ASS CODE

[Sergey Nazarov @sergeynazarovx]

We used to go to a special website, ask strangers for help with programming, and get humiliated in return

[Hater Report @HaterReport]

LeBron James when he sees Deandre Ayton’s ‘DominAyton’ chain

[Kimi.ai @Kimi_Moonshot]

Introducing 𝑨𝒕𝒕𝒆𝒏𝒕𝒊𝒐𝒏 𝑹𝒆𝒔𝒊𝒅𝒖𝒂𝒍𝒔: Rethinking depth-wise aggregation.

Residual connections have long relied on fixed, uniform accumulation. Inspired by the duality of time and depth, we introduce Attention Residuals, replacing standard depth-wise recurrence with learned, input-dependent attention over preceding layers.

🔹 Enables networks to selectively retrieve past representations, naturally mitigating dilution and hidden-state growth.
🔹 Introduces Block AttnRes, partitioning layers into compressed blocks to make cross-layer attention practical at scale.
🔹 Serves as an efficient drop-in replacement, demonstrating a 1.25x compute advantage with negligible (<2%) inference latency overhead.
🔹 Validated on the Kimi Linear architecture (48B total, 3B activated parameters), delivering consistent downstream performance gains.

🔗Full report:
https://t.co/u3EHICG05h

[Alexandria Ocasio-Cortez @AOC]

Why does this guy always talk like a World of Warcraft npc

[Will @WillemDaFro]

Cool man, my plan is also very simple: if an ICE squad decides to bust down my or my neighbors door bc they really wanna cosplay their favorite germans from the 30s and 40s, im running the K/D ratio up as high as I can