Large language models aren't always more complex

Research suggests that large language models don't necessarily perform better than their smaller, better-tuned counterparts.