Picture for Huangqi Yu

Huangqi Yu

AKVQ-VL: Attention-Aware KV Cache Adaptive 2-Bit Quantization for Vision-Language Models

Add code
Jan 25, 2025
Viaarxiv icon