AI RESEARCH
HaLoRA: Hardware-aware Low-Rank Adaptation for Large Language Models Based on Hybrid Compute-in-Memory Architecture
arXiv CS.CL
•
ArXi:2502.19747v3 Announce Type: replace Low-rank adaptation (LoRA) is a predominant parameter-efficient finetuning method for adapting large language models (LLMs) to downstream tasks. Meanwhile, Compute-in-Memory (CIM) architectures nstrate superior energy efficiency due to their array-level parallel in-memory computing designs.