CAP-TTA triggers context-aware preconditioned LoRA updates on high bias-risk OOD prompts to reduce toxicity in LLM narrative generation while preserving fluency and avoiding catastrophic forgetting.
InProceedings of the 60th Annual Meeting of the Association for Computational Linguistics (V olume 1: Long Papers), pages 1878–1893
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Preconditioned Test-Time Adaptation for Out-of-Distribution Debiasing in Narrative Generation
CAP-TTA triggers context-aware preconditioned LoRA updates on high bias-risk OOD prompts to reduce toxicity in LLM narrative generation while preserving fluency and avoiding catastrophic forgetting.