M1 Mac使用conda-env创建TensorFlow虚拟环境失败，报错“An unexpected error has occurred.”

Let’s walk through the most likely issues and fixes based on the diagnostic report you shared:

1. Fix Duplicate PATH Entries #

Looking at your environment variables, your PATH has duplicate entries for /Users/andrewwalker/miniforge3/bin — this can cause conflicts with how Conda locates its own executables. To fix this:

• Open your shell config file (either ~/.zshrc or ~/.bashrc; M1 Macs use zsh by default)
• Find the lines that add Miniforge to your PATH and remove the duplicate entry
• Run source ~/.zshrc (or your corresponding config file) to apply the changes

2. Resolve Conflicting Condarc Files #

Your report shows two populated config files: /Users/andrewwalker/miniforge3/.condarc and /Users/andrewwalker/.condarc. Having two configs can lead to conflicting settings (like channel priorities or package constraints) that break environment creation. Try this:

• Back up your user-level config with mv ~/.condarc ~/.condarc.backup
• Retry running conda env create --file=/Users/andrewwalker/Desktop/tensorconfig.yml --name=arm64test
• If this works, you can merge the settings from the backup into the base config later, checking for conflicts

3. Update Conda to a Newer Version #

You’re using Conda 4.10.3, which is a relatively old version — early Conda releases had limited support for M1’s arm64 architecture. Updating to the latest stable version often resolves these generic errors:

• Run conda update -n base conda in your base environment
• Wait for the update to complete, then try creating your environment again

4. Validate Your YAML File #

A malformed tensorconfig.yml can trigger generic errors even if it looks correct. Test with a minimal, proven YAML first to rule this out:

• Create a test file test_env.yml with:

  name: arm64test
  channels:
    - conda-forge
  dependencies:
    - python=3.9
  

• Run conda env create --file=test_env.yml
• If this works, the issue is in your original tensorconfig.yml — check for syntax errors (like incorrect indentation) or package names that aren’t available for arm64

5. Check Directory Permissions #

While your base environment is marked as writable, permissions issues in your user’s Conda directories can still cause failures:

• Run ls -la ~/.conda to check ownership
• If you see permissions that don’t belong to your user, fix them with chown -R $(whoami) ~/.conda

Start with these steps — the most probable fixes are updating Conda or resolving the conflicting config files. If none of these work, sharing the content of your tensorconfig.yml would help narrow things down further.

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