ej

ej is an innovative open-source AI model optimized for rapid, decision-focused applications. Designed to operate as a compact 11MB local model, it excels at providing probability distributions for decision questions based on given states, all in a single pass. Unlike traditional classifiers, ej handles multiple question types—including choice, yes/no, and ordered scores—making it versatile for various decision-making scenarios. Its lightweight design allows it to run efficiently on CPUs without requiring token generation or extensive computational resources. The tool is particularly suited for developers seeking an embedded solution for real-time decision support, with a focus on flexibility and ease of integration. The inclusion of training, evaluation, adaptation scripts, and benchmarking tools in its repository makes it accessible for customization and performance testing, fostering a community-driven approach to decision modeling.