InfoComp learns per-task private prompts and one shared prompt with parameter-alignment and SimSiam-style losses, reporting consistent accuracy gains over prior state-of-the-art continual text classification methods on 5- and 15-task benchmarks.
Robins, Catastrophic forgetting, rehearsal and pseudorehearsal, Con- nection Science 7 (2) (1995) 123–146
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
Information-Theoretic Complementary Prompts for Improved Continual Text Classification
InfoComp learns per-task private prompts and one shared prompt with parameter-alignment and SimSiam-style losses, reporting consistent accuracy gains over prior state-of-the-art continual text classification methods on 5- and 15-task benchmarks.