Recently updated channels

Pick one to see only that channelBrowse all channels →
MBMatthew Bermantoday +221The Builderstoday +120VC with Harry Stebbingsyesterday +316JSJane Streetyesterday +119AI Engineeryesterday +1硅谷1012 days agoPragerU3 days agoY Combinator3 days agoa16z4 days ago +1SemiAnalysis4 days ago

Results

1 articles
Sort
PersonGrant SandersonLengthOver 30 min
7 slides1:16:29EN
JSJane Street· Aug 26, 2026Grant Sanderson and Alok Puranik: Positional Encodings Through Group TheoryGrant Sanderson, Alok Puranik · 3Blue1BrownStarting with assumptions about attention, they derive matrix exponentials and connect rotations, decay and Jordan blocks to RoPE and ALiBi.Transformers & attentionMathematicsML fundamentals+1