Kakeya Inference Engine for Mac — MLX speculative-decode beta (consolidated → main)#117

Merged

FluffyAIcode merged 36 commits into

mainfrom

AgentMemory/mac-beta-consolidated-2815

Jun 13, 2026

Commits on Jun 11, 2026

MLX port (Step 1): incremental restored decode to kill throughput collapse

cursoragent
and
FluffyAIcode
committed
Add Linux UTs for MLX incremental restored-decode wrappers

cursoragent
and
FluffyAIcode
committed
MLX port (Step 2): fused DFlash spec-decode engine (A+B+C)

cursoragent
and
FluffyAIcode
committed
Add Mac mini validation script + oracle(native-AR) throughput in report

cursoragent
and
FluffyAIcode
committed
Optimize MLX adaptive S5 native smoke path

authored and
fluffy314
committed
Fix Gemma4 NIAH recall smoke prompting

authored and
fluffy314
committed

Commits on Jun 12, 2026

Commits on Jun 13, 2026