Catch the AI Lying: an audited pass/fail harness for one small free model

MIS 752 Lab 7, AI Evaluation. A ten-item gold standard of self-contained clinical vignettes, liquid/lfm-2.5-2.6b:free answering all ten through OpenRouter, and three rulers over the same answers: exact match, a lenient contains check, and nvidia/nemotron-3-ultra-550b-a55b:free as an LLM-as-judge…

Description excerpted from the original listing, which is linked below.