CUDA-parity rollback for the all-MLX fused loop (keep accepted K/V, trim only rejected) — +33% on code, ~AR parity#115

FluffyAIcode wants to merge 4 commits into

AgentMemory/k3-all-mlx-drafter-b876FluffyAIcode/Kakeya-LLM-Inference-engine:AgentMemory/k3-all-mlx-drafter-b876from

AgentMemory/mac-niah-accept-compare-2815FluffyAIcode/Kakeya-LLM-Inference-engine:AgentMemory/mac-niah-accept-compare-2815Copy head branch name to clipboard