L2P (Parameter-efficient fine-tuning (LoRA family)): superseded — cited as a baseline and beaten by newer methods. 1 paper(s) critique it, 7 beat it on benchmarks — #31 of 1113 most-superseded. Sub-problem: cluster led by InfLoRA. Newer alternatives in the same sub-problem include G2LoRA, CoDyRA, Janus-LoRA, E^2-LoRA, Beyond Feature Fusion.

Parameter-efficient fine-tuning (LoRA family)

L2P

Superseded baseline#31 of 1,113 most-superseded

Superseded — cited as a baseline and beaten by newer methods

1 papers critique it · 7 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites L2P as a baseline.

a training trick is employed to mask out the classes not relevant to the current task. This trick contradicts the task-free OCL setting

— Online-LoRA: Task-free Online Continual Learning via Low Rank Adaptation

Beaten on benchmarks

Head-to-head results where a newer method reports beating L2P. Values are copied from the source paper's tables — verify against the cited paper.

E²-LoRA beats L2P
75.82 vs 38.18
Last-Acc · [Cars-196 10-task]
Energy-Structured Low-Rank Adaptation for Continual Learning
DualLoRA+ beats L2P
1.10 vs 5.36
FT · [5-Split ImageNet-R]
Replay-Free Continual Low-Rank Adaptation with Dynamic Memory
SD-LoRA beats L2P
35.12 vs 70.14
GFLOPs · [ImageNet-R (N=20)]
SD-LoRA: Scalable Decoupled Low-Rank Adaptation for Class Incremental Learning
SDLoRA beats L2P
55.96 vs 42.94
Acc · [ImageNet-A (N=10)]
SD-LoRA: Scalable Decoupled Low-Rank Adaptation for Class Incremental Learning
CoDyRA beats L2P
69.1 vs 53.2
Average · [Transfer]
Take Only What You Need: Rank Minimization as an Implicit Forgetting Regularizer in Continual Learning
CL-LoRA beats L2P
81.58 vs 65.82
Average Accuracy · [ImageNet-R T=40]
CL-LoRA: Continual Low-Rank Adaptation for Rehearsal-Free Class-Incremental Learning
InfLoRA beats L2P
71.01 vs 57.92
ACC_20 · [20 tasks]
InfLoRA: Interference-Free Low-Rank Adaptation for Continual Learning
L2L beats L2P
57.87 vs 47.91
Average Accuracy · [memory-free continual learning]
Choice of PEFT Technique in Continual Learning: Prompt Tuning is Not All You Need

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.