LM Compass is a web-based platform designed to evaluate and rank the quality of responses to queries from multiple Large and Small Language Models (LLMs/SLMs) using a scalable LLM-as-a-judge cross-evaluation system. By integrating user customizable judging rubrics and a variety of evaluation methods, we built an open and unbiased framework for researchers and developers to determine which models are best suited for specific queries or large sets of data.
Developed by Sohaib Ahmed, Aryan Suvarna, Aadi Sanghani, Gulkaran Singh, Madhav Kalia, Owen Jackson & Rochan Muralitharan under supervision of Dr. Angela Zavaleta Bernuy
…Read more
Less…