LQ-LoRA: Low-rank plus Quantized Matrix Decomposition for Efficient Language Model Finetuning

Han Guo, Philip Greengard, Eric P. Xing, Yoon Kim. LQ-LoRA: Low-rank plus Quantized Matrix Decomposition for Efficient Language Model Finetuning. In The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net, 2024. [doi]

Abstract

Abstract is missing.

  翻译: