Fetching the paper…

Procedural Knowledge in Pretraining Drives Reasoning in Large Language Models · Around