Asking opus or fable to perform tests on a large dataset saves me a lot of time. But if i ask it to go back through and doublecheck those results it will often tell me they’re bullshit.
6
Don’t piss in my pocket and tell me it’s raining sec.gov/Archives/edgar/data/…
14
So covid may have come from a virology institute. This plague may have come from a plague institute. Im seeing a pretty clear solution here.
1
33
Sorry but Claude Opus 5.5 just suggested a train test split by date where the train split used > date with 5% of total samples. THAT IS MADNESS. It’s like Stephen Hawking with a massive lobotomy.
1
44
The AI Bubble is a bubble, it’s gonna pop @ end of 2027. Attention Is All You Need, amazing paper but capital bloated morons pushed too much air into a blimp, now it’s falling. AI was so much fun in 2020 and now it’s boring. Get out of life insurance if you have a policy.
31
The only people that don’t mind TruMotion are the ones that don’t know how to run it off. This has to stop. How does anybody enjoy this? @SamsungUS @SonyElectronics @LGUS
29
Building power infrastructure for both EVs AND AI would lower costs for everyone. Instead we're subsidizing one and kneecapping the other. Why? #Stargate
32
Unlocking data quality through Large Language Models (LLMs). Provide an initial prompt, then loop with a second prompt to traverse knowledge landscapes & generate high-quality datasets @ollama
1
25
Prompt 1: Identify a fundamental concept in your knowledge base. Explain its core principles, then examine an unexpected or non-obvious implication of this concept. What new questions or areas of inquiry does this implication open up?
1
22





