Search

This paper tackles the challenge of building robust and generalizable bias mitigation models for language. Recognizing the limitations of existing datasets, we introduce ANUBIS, a novel dataset with 1507 carefully curated sentence pairs encompassing nine social bias categories. We evaluate state-of-the-art models like T5, utilizing Supervised Fine-Tuning (SFT), Reinforcement Learning (PPO, DPO), and In-Context Learning (ICL) for effective bias mitigation. Our analysis focuses on multi-class social bias reduction, cross-dataset generalizability, and environmental impact of the trained models. ANUBIS and our findings offer valuable resources for building more equitable AI systems and contribute to the development of responsible and unbiased technologies with broad societal impact., Comment: 17 pages, 5 Figures

The Generational Garbage collection involves organizing the heap into different divisions of memory space in-order to filter long-lived objects from short-lived objects through moving the surviving object of each generation Garbage Collection cycle to another memory space updating its age and reclaiming space from the dead ones. The problem in this method is that the longer an object is alive during its initial generations the longer the garbage collector will have to deal with it by checking for its reachability from the root and promoting it to other space divisions where as the ultimate goal of the Garbage Collector is to reclaim memory from unreachable objects at a minimal time possible. This paper is a proposal of a method where the lifetime of every object getting into the heap will be predicted and will be placed in heap accordingly for the garbage collector to deal more with reclaiming space from dead objects and less in promoting the live ones to the higher level.