mel

growing on twitter is genuinely just luck

@lstmytouch and i'm the unluckiest mf on earth

Growing on twitter is genuinely just luck

@1a1n1d1y llm inference is more important

@arm64le give it to someone in need

the only thing between me and asi is this disc

A pretty critical evolutionary event about 400,000 years ago??

@vSouthvPawv it should work out of the box!

Hermes is the agent that grows with you, even if you do not grow up yourself 😅

It’s finnnnnnnally out: huggingface.co/moonshotai/Kim…

KIMI K2.6 MY DEARLY BELOVED

Kimi-K2.6 Benchmarks

@scaling01 always bet on moonshot

Kimi-K2.6 SWE-Bench-Pro results are ridiculous

BREAKING: Google DeepMind has assembled a strike team because Anthropic is mogging them on coding Led by Sergey Brin and DeepMind CTO Goal: Force recursive self-improvement by turning coding models into full AI researchers that can automate the entire R&D loop GDM is focusing on: >long-context coding tasks >training models on GDM’s private codebase “To win the final sprint, we must urgently bridge the gap in agentic execution and turn our models into primary developers” ACCELERATE

got so distracted from the stuff i should be working on just to bully a python user

@miaaowing the people yearn to be uneducated

how are so many people fundamentally uninterested in how the things they use every day function. like at all

i keep mixing up llvm and vllm when saying stuff out loud gg

why do we not have uint_24

I am sick of people crediting "test time compute" to OpenAI. Deepseek zero was published much before o1 was in the works. It was deepseek. It wasn't OpenAI. Stop trying to rewrite history

@yacineMTB sloppy RL got sloppier, plus their embedding dims are probably like half a bit at this point with how much it regressed on NIHS

No one I know likes opus 4.7 How could they mess it up so badly?

1) skill 2) competence 3) understanding of fundamentals 4) none of the below

agi agi agi agi agi agi agi

night city coded