Tera Luna was a ponzi from day 1 🐻

gm

Rapidly rising cost to train LLMs.
$10 million to train GPT-3.
$100 million for GPT-4?
🐻

GPT-3’s forced alignment, thanks to OpenAI’s latest alignment retraining.
🐻

GPT-3's big political realignment.

Will China politically-realign theirs too? NK? Iran?

Brings a whole new meaning to linguistic determinism.

https://davidrozado.substack.com/p/the-political-orientation-of-the

🐻