The Sharpness Disparity Principle in Transformers for Accelerating Language Model Pre-Training

Re-assign community