DeepSeek-V3.2: “Intelligence will become too cheap to meter” - YouTube
https://www.youtube.com/watch?v=pljoUcBniPQ
Breaking down what made DeepSeek V3.2 such an important paper, how is DeepSeek-V3.2-Speciale so good, how DeepSeek has created this model, and explaining DeepSeek's new secret weapon: DeepSeek Sparse Attention (DSA).
https://stacker.news/items/1320313