DL-QAT: Weight-Decomposed Low-Rank Quantization-Aware Training for Large Language Models.

Wenjing Ke, Zhe Li, Dong Li, Lu Tian, Emad Barsoum

Browse the full EMNLP paper archive.