Taming LLMs by Scaling Learning Rates with Gradient Grouping

Re-assign community