{"blocks": [{"key": "69640014", "text": "Question", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}, {"key": "10b64c28", "text": "In Machine Learning, what are the high-level trends happening at the framework level? How are frameworks evolving from NumPy to PyTorch to JAX, and what are three key differences between PyTorch and JAX? What are the stages a model goes through from being defined to running on a GPU? Describe the typical frontend, intermediate representation (e.g., ONNX computation graph), and compilation steps. What optimization techniques are applied during model compilation for GPUs? Discuss kernel fusion, quantization, and other relevant methods. Are you familiar with data-center hardware versus edge hardware, and how does that influence compilation or deployment choices?", "type": "unstyled", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {}}], "entityMap": {}}