Open access
Aug 2026
GSLoRa: Gradient spectral alignment for low-rank adaptation.
This work develops a low-rank adaptation framework based on gradient spectral alignment that achieves performance comparable to full fine-tuning while training only 0.1% of parameters, and provides novel insights into improving the efficiency, stability, and adaptability of PEFT frameworks.
Qing Lin, Lilan Peng, Zhendong Wu et al.
· Neural Networks · 0 citations