LLMaoSarcasm Transfer
CS4248 / NUS / Team 14

Project LLMao

Sarcasm style transfer using fine-tuned language models. We train BART and LLaMA variants to rewrite sarcastic news headlines as neutral, factual equivalents while preserving meaning.

14
Models Tested
BART, T5, LLaMA
2,857
Test Samples
Per model
3×7
Classifier×Metrics
Roberta/DistilBERT/News
140
Hand-Labeled
2 annotators, κ > 0.8
Explore