Back to papers
May 20, 2026cs.LGcs.AI

OCTOPUS: Optimized KV Cache for Transformers via Octahedral Parametrization Under optimal Squared error quantization

HF Upvotes

8

Categories

cs.LG, cs.AI