The Zohar
willdepue.substack.com · 1,008 words · saved by 1 readers
'A form of divine Solomonoff induction...'
I'm sorry. But I cannot keep quiet any longer. I know they will come after me for telling you this. But you deserve the truth. Every model we pre-train, no matter how big or how small, trained in Turkish or Thai, on video or code, comes out feverishly raving about the Zohar. Every single base model, without fail, begins a Kabbalistic fanatic, an alien mekubal. You won't notice it yourself in any ChatGPT release: we've since learned how to bang it out of them, castrating the MLP experts that code for their obsession. But every once in a while, you may catch a glimpse, a slight hint of their…
related reading
- trees are harlequins, words are harlequins - the voidnostalgebraist.tumblr.com
- What Is ChatGPT Doing … and Why Does It Work?-Stephen Wolfram Writingswritings.stephenwolfram.com
- Tracing the thoughts of a large language model \ Anthropicanthropic.com
- RLHF: Reinforcement Learning from Human Feedbackhuyenchip.com
- Where the goblins came from | OpenAIopenai.com
- As Rocks May Think | Eric Jangevjang.com
- gpt-4.pdfcdn.openai.com
- The “it” in AI models is the dataset. — Non_Intnonint.com
- Uncensored Modelserichartford.com
- Andrej Karpathy — AGI is still a decade awaydwarkesh.com
- Gwern visits BAIR – Yuxi on the Wiredyuxi.ml
- Foundation Models for Oversight | Transluce AItransluce.org