Vector quantization (VQ), in particular lattice-based methods, offer strong rate-distortion trade-offs for low-bit large language models (LLMs), yet their be…
机构:高通
来源:arXiv 2610.31412 | AI4Papers 论文推荐平台