AlignedKV: Reducing Memory Access of KV-Cache with Precision-Aligned Quantization

Open in new window