NVIDIA Minitron: Compact Language Models via Pruning and Knowledge Distillation

知识蒸馏 / 模型剪枝