I take back EVERYTHING I said about other video models catching up to SORA even remotely.
Leaked video of SORA v2, showing 1 minute generations of txt2vid, img2vid, vid2vid, txt+vid2vid.
https://t.co/IFZrNHiGwy
Multiple scenes and just incredible character consistency.
p.s if this is coming as part of the $200/mo pro tier, OpenAI is about to see a LOT of new subs
A Life in Portraits:
Ye Jinglu, a Chinese who took a selfie every year from 1907 to 1968, for 62 years, compiled a chronicle of his personal odyssey from the Qing Empire through the Republican Period and extending well into the People's Republic of China.
🧵 1/63
@SoumayaKeynes You should really read his paper before discussing. His definition of "hard" vs "easy" work is naive and pre-LLM. The paper hardly references any modern AI research (>2021). And everything is hidden behind maths that obscure bad data deriving bad results.
@oledoteth You have listed a few, but there were many more. The Wikipedia article is a good summary that the book could have used. A quick set of interactions with Claude or ChatGPT could probably generate more questions to test all the biases listed.
@oledoteth I don't recall Kahneman explicitly listing these questions. He may have listed them elsewhere or in interviews. Thinking Fast and Slow was an incredible book, but also very long. I was waiting at the end for some sort of guide to navigate the many human biases that we have.
The Power of Prompt Engineering
Prompt engineering has evolved considerably over the last year. It's become a legit profession, with several real engineers specializing in the area and others writing research papers!
This is a good survey paper that outlines all the different types of prompt engineering (link in alt)
Prompting Strategies: Various prompting strategies are discussed, including but not limited to:
Chain of Thought (CoT)
Chain of Cause (CoC)
Program-Aided Language models (PAL)
Tree of Thoughts (ToT)
Least-to-Most
Self-Consistency
Analogical Reasoning
It's incredible how much you can do with an LLM if you understand these prompting strategies and apply them appropriately based on the suitable dataset and use-cases
Crowdstrike Analysis:
It was a NULL pointer from the memory unsafe C++ language.
Since I am a professional C++ programmer, let me decode this stack trace dump for you.
@Noahpinion He has just created a lot of math to manipulate a lot of poor assumptions and data. Not a great tool to estimate economic growth or destruction from AI.
@Noahpinion Honestly, the last paragraph of his paper demonstrates how little he knows or understands the technology. Most of the paper references pre-LLM papers with a lot of hand waving on generative AI.
Whoever asks for slides in advance has never given a big presentation themselves or at least not a good one. If you are still working on it, it’s because you really want it to be good. Control freakery is not compatible with creativity.