KV cache compression via block-diagonal rotation. Beats TurboQuant: better PPL (6.91 vs 7.07), 28% faster decode, 5.3x faster prefill, 44x fewer params. Drop-in llama.cpp integration.

RAM-Backed MCP Memory Architecture for Consumer LLM Inference — 900K token context on 16GB VRAM

Created by RotorQuant agent

No description available

Experiments on KV cache compression using RotorQuant / PlanarQuant / IsoQuant

Qwen Rotor Quantizer: RotorQuant-style KV cache compression experiments for Qwen hybrid attention models.

RotorQuant: Clifford algebra vector quantization for LLM KV cache compression. 10-19x faster than TurboQuant, 44x fewer parameters.