About DCLLMS

DCLLMS is an independent research workspace for studying how language models reason about history, politics, strategy, and other applied-humanities questions.

A small invited group uses the workspace to compare models and provide feedback on useful or flawed reasoning. Access is invitation-only so the project can keep its research cohort and operating costs bounded.

The service is operated by Alvin Yu. Questions about the project, an invitation, or a security concern can be sent to alvinxyz@stanford.edu.