http://markbhall.dev/Recent content on Mark HallHugoen-USMark HallMon, 27 Jul 2026 22:25:45 -0600http://markbhall.dev/writing/my-local-llm-scored-6-of-6/Mon, 27 Jul 2026 22:25:45 -0600http://markbhall.dev/writing/my-local-llm-scored-6-of-6/<p>One of the tasks in my benchmark asks this:</p> <blockquote> <p>How many 5-person committees can be chosen from 12 people? Reply with only the integer.</p> </blockquote> <p>My 1.2B local model answered <code>60</code>. The correct answer is 792.</p> <p>My evaluator passed it.</p>