Method Drift›LLM reasoning / chain-of-thought
LiDER
LLM reasoning / chain-of-thought
superseded — cited as a baseline and beaten by newer methods
1 papers critique it · 1 beat it on benchmarks
What papers say
Verbatim critique sentences, each from a paper that cites LiDER as a baseline.
“LiDER only considers model Lipschitz constants which can be implicitly related to the landscape of model outputs in a local region but does not consider the input data”
— STAR: Stability-Inducing Weight Perturbation for Continual Learning
Beaten on benchmarks
Head-to-head results where a newer method reports beating LiDER. Values are copied from the source paper's tables — verify against the cited paper.
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+ER-ACE beats LiDER · accuracy [ER-ACE + buffer size 100]
60.69 vs 56.08
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+ER-ACE beats LiDER · accuracy [ER-ACE + buffer size 200]
67.58 vs 65.32
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+ER-ACE beats LiDER · accuracy [ER-ACE + buffer size 500]
75.44 vs 71.75
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+ER-ACE beats LiDER · accuracy [ER-ACE + buffer size 2000]
51.67 vs 50.32
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+ER-ACE beats LiDER · accuracy [ER-ACE + buffer size 1000]
21.06 vs 19.69
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+ER-ACE beats LiDER · accuracy [ER-ACE + buffer size 5000]
31.01 vs 30.00
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+DER++ beats LiDER · accuracy [DER++ + buffer size 100]
61.76 vs 58.43
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+DER++ beats LiDER · accuracy [DER++ + buffer size 200]
68.60 vs 66.02
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+DER++ beats LiDER · accuracy [DER++ + buffer size 500]
76.52 vs 73.39
- STAR: Stability-Inducing Weight Perturbation for Continual Learning
STAR+DER++ beats LiDER · accuracy [DER++ + buffer size 1000]
22.4 vs 21.58