Antialpha

[Joe Doliner @jdoliner]

Google could discover a new architecture that beats the transformer and they would still find a way to lose the ai race

[Saket Gokhale @SaketGokhale]

@JagadeeshPuttam The award citation was posted by the official handle of Modi’s party BJP. Do you have ANY shame at all?

[Pimpjuice💤 @RealGame101]

If you were on a plane, and the pilot says on the loudspeaker

“Hey guys, just so you know, I’m feeling really low and I don’t want to be alive anymore”

Would you swap the pilot out with a stable one if given the option?

1000%

This is how girls view your mental health issues

[Could We Not? @DecidedlyAngry]

Decades ago, after years of hearing about how I needed to be more in touch with my emotions, when I was really upset about something traumatic, I finally did exactly what she asked me to do. She got what they now refer to as "the ick" immediately. Like, immediately. It was all over her face. She never looked at me the same since. Conversely, I've never looked at her the same since..

Don't believe them, fellas. What they say they want and what they really want aren't in the same universe. It's a trap.

[Maririn~ @TopGyaru]

Scary

[Stix @ByteStix]

@TopGyaru JUMPSCARE

[Sanju Choudhary @saaaanjjjuuu]

BREAKING : Samsung, SK Hynix & Micron are facing a class action lawsuit in the US as memory prices are out of control due to shortages!!

These tech giants are getting sued for intentionally choking the supply of RAM while demand was exploding. They knew exactly what they were doing, and it triggered the infamous “RAMpocalypse” where memory prices went to the absolute moon.

The lawsuit seeks action for damages on behalf of consumers who allegedly paid inflated prices for products containing DRAM memory.

[Antialpha @Antialpha8]

@DrKewp @ai_sentience @scaling01 What made you think that machines are not vulnerable or frail? Did you even think your statement through?

[青龍聖者 @bdsqlsz]

A habit developed seven years ago.

[Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex]

Good guy DeepSeek gives us accelerated models
The most interesting one here is Gemma4-12B, I presume vision included. Might be the best local model in its weight class now, by some margin
Qwen 3.5 not included because DS[park] doesn't do linear attention I guess

[Antialpha @Antialpha8]

@Grace_rinchan Chatgpt and nsfw never worked for me

[Elon Musk @elonmusk]

To be clear, I’m not saying the Grok v9 foundation model will be mind-blowingly better than anything, but it will be a solid workhorse in the same league as Opus.

And the SpaceXAI cadence of model and harness improvement is speeding up tremendously, particularly due to a few dozen of the top Starlink/Starship engineers shifting a much of their time to AI.

Worth noting that the v8 foundation (Grok 4.3) model is a 0.5T model that finished training all the way back in December and has many fundamental flaws, so Grok 4.5 will feel like a gigantic upgrade!

[Antialpha @Antialpha8]

@mark_k @GoogleDeepMind They implemented token based usage limits with 5 hr windows. It definitely took a hit after that but free and low tier users are getting more usage now.

[Antialpha @Antialpha8]

@DrKewp @ai_sentience @scaling01 why does biology matter? What makes biochemistry superior to mechanics if in the end it's all based on chemistry, physics and math?

[Skyler @skylermzx]

this shouldnt be allowed.
why should the government dictate what i can install on my phone?

[adi @adonis_singh]

I doubt china even has a 4.8 competitor, let alone mythos

[Lisan al Gaib @scaling01]

Sounds like we will get the 1.5T model in July and the 2T model in August

[Elon Musk @elonmusk]

@scaling01 Every aspect of the 2T model is improved over the 1.5T, so it will be much more than just a 1/3 increase in params

[GDP @bookwormengr]

DeepSeek invents trick to save on HBM bandwidth, or even raw memory bits.

Memory bag holder: let me explain you how it is going increase demand of memory 10x 🚀.

🤦‍♂️

[Dr Nilima Srivastava @gypsy_nilima]

@AroonDeep That’s why the error in spelling of Republic

Spelt with two BB’s