《深度学习基础》TensorFlow脚本导入read_data模块报错求助

Hey there! Let's tackle this problem you're facing with the TensorFlow script from Fundamentals of Deep Learning by Nikhil Buduma.

First off, the read_data module isn't a built-in part of TensorFlow or standard Python libraries—this is almost certainly a custom helper module the author created for the book's examples, meant to handle tasks like loading datasets or generating minibatches (which ties into the get_minibatch function you already adjusted).

Here are your best options to resolve this:

• Grab the book's official companion code: Most technical books host their example code on a repository (like GitHub). Look for the book's official codebase—you'll find the read_data.py file there. Once you have it, place it in the same directory as your running script, or add its folder path to Python's sys.path so your script can locate it.
• Implement the functionality yourself: If you can't find the official code, you can recreate the key parts. Since you're working with get_minibatch, this function likely pulls small batches of data from a dataset. Depending on the dataset you're using (e.g., MNIST), you can use TensorFlow's native tools like tf.data.Dataset to handle batching, or write a simple function to split your data into minibatches manually.
• Double-check for typos: It's easy to misspell module names—double-check the book's code example to make sure you're referencing the module exactly as written (e.g., is it read_data or something like read_data_utils?).

Also, quick note: You were right to remove the empty parentheses from get_minibatch! If the script was passing the function as a parameter (common in TensorFlow workflows), adding parentheses would execute the function immediately instead of passing the function itself, which would cause an error. Good catch there.

内容的提问来源于stack exchange，提问作者DarkHark