LQ-LoRA: Low-rank plus Quantized Matrix Decomposition for Efficient Language Model Finetuning.
Han GuoPhilip GreengardEric P. XingYoon KimPublished in: ICLR (2024)
Keyphrases
- low rank
- matrix decomposition
- language model
- low rank matrix
- linear combination
- nonnegative matrix factorization
- missing data
- matrix factorization
- singular value decomposition
- convex optimization
- kernel matrix
- singular values
- high order
- probabilistic model
- high dimensional data
- semi supervised
- data matrix
- latent structure
- data sets
- information retrieval
- image classification
- negative matrix factorization
- pairwise
- feature selection
- data summarization