Hey there! Let’s tackle your Miniconda naming confusion step by step—this is a super common question, so you’re not alone.

1. Miniconda-3.6.0-Windows-x86_64.exe vs Miniconda3-3.6.0-Windows-x86_64.exe 的区别 #

The key here is the 3 after Miniconda:

• Miniconda-xxx (no trailing 3): This is the legacy line of Miniconda that defaults to Python 2 environments. Even though you could manually create a Python 3.6 environment with it, the base environment will be Python 2, which isn’t what you want for a TensorFlow setup that requires Python 3.x.
• Miniconda3-xxx (with trailing 3): This is the line designed for Python 3 as the default. The 3.6.0 here directly refers to the default Python version pre-installed in the base environment. So this is the one you should pick for your Python 3.6 + TensorFlow setup.

2. Miniconda的完整命名规则 #

Let’s break down a typical filename like Miniconda3-4.5.12-Windows-x86_64.exe:

• Miniconda3: Indicates this is the Python 3-focused distribution (as opposed to the Python 2-focused Miniconda).
• 4.5.12: This is the version number of Miniconda itself, not Python! Miniconda is a bundle that includes conda, pip, and a default Python version. This version tracks updates to the conda package manager and other bundled tools, not the Python version. To find out which Python version comes pre-installed with a specific Miniconda release, you’ll need to check the release notes or the archive page’s accompanying descriptions.
• Windows: Target operating system.
• x86_64: CPU architecture (64-bit; x86 would be 32-bit).

So for your use case: stick with Miniconda3-3.6.0-Windows-x86_64.exe—it’s purpose-built to give you a Python 3.6 base environment out of the box, which aligns perfectly with TensorFlow’s Python 3.x requirements.

内容的提问来源于stack exchange，提问作者HelloGoodbye