Model Interpretability for PyTorch. ... If any of the tuple elements is a slice object, the indexed output tensor is used for attribution.
確定! 回上一頁