Evaluate performance of LLM models for Q&A in any domain

Automated search framework for rubric-based reward modeling. Features Evolutionary RTD (population search with elitism + successive halving) and Iterative RTD baseline. Supports tail-focused objectives, multi-role LLM backends, and rank-based preferences

AI-powered student assignment evaluator written in Rust. Supports code, PDF, and DOCX files. Uses local or remote LLMs to grade submissions based on configurable criteria, and exports results to Excel.