Don't Drop Dropout： Optimizing Layer Sparsity for Efficient LLM Training and Inference

This record preserves its title, category and source. Open the original source for full details.