LID-ABS: Stress-testing Machine Generated Text Detection: Shifting Language Models Writing Style to Fool Detectors
Datasets generated by aligning LLMs using Direct Preference Optimization to shift the machine-generated texts' (MGT) style toward human-written text (HWT).
This dataset is intended to be used to augment the training set of documents to train more robust MGT detectors, or to evalaute the robustness of MGT detection systems.
