PyTorch 实现与追问

题集来源:Alireza Dirafzoon(alirezadir)收录项目:AIMLInterviews查看原题集

题目来源
Devinterview 其中第 85—146 题的题目索引来自 Devinterview,答案由 AIMLInterviews 重新撰写;原文的 50 题补充部分在此收录为 62 道,含拆分追问。

题目目录

156 道原题
  1. Support classification and regression in one loop.

    原题
  2. Add callback hooks cleanly.

    原题
  3. Move nested batches to a device.

    原题
  4. Preserve input order and sample identifiers during prediction.

    原题
  5. Test a standardization utility, including constant features.

    原题
  6. Verify evaluation does not update model state.

    原题
  7. Compare CPU and GPU metric implementations.

    原题
  8. Test empty input, batch size one, and unexpected shapes.

    原题
  9. Use a tiny dataset to verify overfitting.

    原题
  10. Diagnose a loss that never changes.

    原题