Training large language models (LLMs) in sub-8-bit arithmetic promises substantial gains in memory bandwidth and compute throughput, yet FP4 formats introduc…
机构:AMD
来源:arXiv 2610.32229 | AI4Papers 论文推荐平台