Abstracts: January 25, 2024

Abstracts brings you the latest achievements and breakthroughs from Microsoft's research community. In this episode, Senior Researchers Jordan Ash and Dipendra Misra discuss their paper "The Truth is in There: Improving Reasoning in Language Models with Layer-Selective Rank Reduction," which was accepted into the International Conference on Learning Representations. The paper introduces LASER, a method for parameter reduction in transformer-based models that can improve model performance without additional training. To learn more about the work and related topics, viewers can register for Microsoft Research Forum, a series of discussions and lightning talks on science and technology research in the era of general AI.

For more information on the paper and code, check out the provided links. Stay up to date with cutting edge research and achievements by tuning in to Abstracts.