Published some notes on Dario Amodei's new essay on DeepSeek, mainly to highlight some new-to-me details he included about Claude 3.5 Sonnet
https://simonwillison.net/2025/Jan/29/on-deepseek-and-export-controls/
@simon re: the cutoff date:I don't fully understand how the (pre) training works, but it seems you might keep updating the "corpus" (pile of documents) while you're training, as long as it's a relatively small portion of the data.
QOTO: Question Others to Teach Ourselves An inclusive, Academic Freedom, instance All cultures welcome. Hate speech and harassment strictly forbidden.