LRQ: Optimizing Post-Training Quantization for Large Language Models by Learning Low-Rank Weight-Scaling Matrices.
Jung Hyun LeeJeonghoon KimJune Yong YangSe Jung KwonEunho YangKang Min YooDongsoo LeePublished in: CoRR (2024)
Keyphrases
- language model
- low rank
- learning algorithm
- matrix completion
- language modeling
- supervised learning
- singular value decomposition
- n gram
- missing data
- low rank matrices
- information retrieval
- low rank matrix
- singular values
- linear combination
- query expansion
- active learning
- higher order
- matrix factorization
- web search
- small number
- pattern recognition
- training data
- feature selection