Daoguang Zan

PanGu-Coder2: Boosting Large Language Models for Code with Ranking Feedback

2023 · arXiv (Cornell University)

Large Language Models for Code (Code LLM) are flourishing. New and powerful models are released on a weekly basis, demonstrating remarkable performance on the code generation task. Various approaches have been proposed to boost the …

The Devil is in the Neurons: Interpreting and Mitigating Social Biases in Pre-trained Language Models

2024 · arXiv (Cornell University)

Pre-trained Language models (PLMs) have been acknowledged to contain harmful information, such as social biases, which may cause negative social impacts or even bring catastrophic results in application. Previous works on this problem mainly focused …