Recently, I got out of googlecolab and wanted to use local anaconda to conduct machine learning exercises. I reported an error when installing tensorflow: UnsatisfiableError: The following specifications were found. The solution is given below.
It is found that the actual reason is due to the python environment of anaconda. The current version of tensorflow can only be applied to derivative versions such as python 3.5, 3.6 and 3.7, while the python version that comes with anaconda is 3.8.
Use the following code to view the current anaconda version and determine whether anaconda is installed correctly.
conda --version
My version here is conda 4.9.1.
Then use the following code to list all the environments in conda:
conda info --envs
Then use the following code to view the python version in the current default environment:
conda activate
python -V
My current version is python 3.8.
In order to successfully install tensorflow, we first need a low-version python environment.
Execute the following command to create an environment named python37, and follow the prompts to install python3.7
conda create --name python37 python=3.7
Entering the new environment, we re-check whether the python version is the 3.7 version we need.
activate python37
python -V
After configuring our python3.7 version, we can install tensorflow in various ways. I use the concise graphical interface of anaconda navigator here.