Turning Logic Against Itself:
Probing Model Defenses Through Contrastive Questions
Jailbreak attack to bypass LLM safety mechanisms.
Turning Logic Against Itself:
Localizing and Mitigating Errors in Long-form Question Answering
Fine-grained error evaluation of long-form LLM responses.
Are Emergent Abilities in Large Language Models just In-Context Learning?
Evaluation of the hype around LLM abilities..
CATfOOD: Counterfactual augmented training for improving out-of-domain performance and calibration
Counterfactual augmentation of data to improve out-of-domain generalization of models.
UKP-SQuARE v2: Explainability and Adversarial Attacks for Trustworthy QA
Platform to interpret and explain the predictions of machine learning models.
UKP-SQUARE: An Online Platform for Question Answering Research
An online QA platform which allows users to query, compare and evaluate different models.