QR-LoRA trains only scalar coefficients on a pivoted QR basis of pretrained weights, reportedly matching full fine-tuning on GLUE with 601 parameters.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
QR-LoRA: QR-Based Low-Rank Adaptation for Efficient Fine-Tuning of Large Language Models
QR-LoRA trains only scalar coefficients on a pivoted QR basis of pretrained weights, reportedly matching full fine-tuning on GLUE with 601 parameters.