As language models (LMs) improve at tasks like image generation, trivia questions, and simple math, you might think that human-like reasoning is around the corner. In reality, they still trail us by a wide margin on…
Enabling small language models to solve complex reasoning tasks | MIT News
