Tech NewsJuly 13, 2024Meta researchers distill System 2 thinking into LLMs, improving performance on complex reasoning