TensorFlow文档中整数值与数值下划线的含义探究：下划线是否等同于小数点？

Hey there! Awesome question—those underscores you're seeing in numbers (like 10_000 for batch sizes) have nothing to do with decimal points. Let's break this down simply:

• It's a Python readability feature: Introduced in Python 3.6, underscores act as digit separators for numeric literals. They’re purely there to make large numbers easier to scan at a glance—think of them like commas you’d use when writing numbers (e.g., 10,000 becomes 10_000 in code).
• No impact on value or type: The Python interpreter completely ignores these underscores. So 10_000 is identical to 10000—same integer value, same data type. A decimal point (.) would change things entirely (e.g., 10.000 is a float equal to 10.0), but underscores don’t alter the number at all.
• Why TensorFlow uses it: You’ll spot this for values like batch sizes, dataset counts, or hyperparameters where large numbers are common. It’s a standard Python best practice now, so it’s used across TensorFlow docs and other Python-based ML codebases to keep code clean and easy to read.

This is just syntax sugar for readability—no hidden functionality here!

内容的提问来源于stack exchange，提问作者sakeesh