Test Run Comparisons

LangChain has introduced Test Run Comparisons to assist developers in building and evaluating Language Model (LLM) applications. The new feature allows users to compare multiple test runs side-by-side, making it easier to gain insights from the data. Users can also use LLM-assisted evaluation or other methods for scoring tests and manually explore datapoints for further analysis. This tool aims to help developers better understand how their LLMs are performing on specific tasks by comparing different test runs on a dataset. LangSmith is currently in private beta, with more access and features expected to be added over the coming weeks.