Optimum + ONNX Runtime：讓 Hugging Face 模型訓練更簡單、更快速

As the scale of deep learning models (such as Transformers) continues to grow, training these models demands enormous computational resources and time. To help developers reduce training costs and improve efficiency, Hugging Face partnered with Microsoft to integrate ONNX Runtime (ORT) training acceleration capabilities into Hugging Face's Optimum toolkit. ONNX Runtime is not only exceptional at model inference — its training module (ORT Training) can also significantly accelerate the training of PyTorch models through various graph optimization techniques and efficient memory management.