Tagged: attention

1 article

May 4, 2026 DeepSeek V4: Don't Look at What You Don't Need DeepSeek V4 reads a million tokens on roughly a quarter of V3.2's compute. It does this by selectively attending to the parts of context the prompt asks about, the same way humans skim a long book.
llmaiattentiondeepseek
Read article
All articles