3 ms·
The papers from Anthropic on interpretability are pretty good. They look at how certain concepts are encoded within the LLM.
by libraryofbabel 1y ago
The papers from Anthropic on interpretability are pretty good. They look at how certain concepts are encoded within the LLM.