LLM internals and the ‘stochastic parrot’ debate
A post argues that looking inside LLMs with mechanistic interpretability could help settle the debate over whether they’re ‘stochastic parrots.’
TLDR
The author claims that even standard LLMs internally rely on structures like those philosophers describe in papers about understanding. They argue that examining those structures could help settle the ‘stochastic parrot’ debate.
Combined views
8.8K
2 Sources, first seen 22h ago
90 likes