Method Drift›Parameter-efficient fine-tuning (LoRA family)
Superseded baseline#239 of 1,113 most-superseded
SWAG-LoRA
Parameter-efficient fine-tuning (LoRA family)
superseded — cited as a baseline and beaten by newer methods
2 papers critique it · 0 beat it on benchmarks
What papers say
Verbatim critique sentences, each from a paper that cites SWAG-LoRA as a baseline.
“the SWAG-LoRA method~onal2024gaussian needs additional computation to obtain an estimation of the posterior.”
— Variational Low-Rank Adaptation Using IVON“One major issue is the higher storage and memory requirements, as Bayesian methods often require additional parameters to model uncertainty, particularly those involving covariance estimation, such as SWAG-LoRA.”
— Minimal Ranks, Maximum Confidence: Parameter-efficient Uncertainty Quantification for LoRA