ByteBrief
We're a portrait publication through and through. Turn your phone back and your briefing picks up right where you left it.
(We tried widescreen once. It wasn't us.)
A 26-billion-parameter Gemma 4 mixture-of-experts model now runs at 5 tokens per second on a 13-year-old Ivy Bridge Xeon with no GPU. The patch fixes a silent fall-through bug in ik_llama.cpp that caused gibberish output on pre-AVX2 chips. The fix splits fused matmul ops into separate working paths.
Tap to vote and see what everyone thinks.
Summary by ByteBrief