目录
背景
真是头疼 windows的anaconda环境这么容易出错...
今天由于在base环境中安装了不同版本的tf(忘了先激活到虚拟环境tf1.15) 导致anaconda有了bug 导入tf、numpy有问题 卸载重装也不行 然后conda activate命令用不了 按网上anaconda回滚的操作 结果直接conda命令也用不了了
卸载重装Anaconda3-5.2.0-Windows-x86_64 装到了D盘 但是jupyter notebook打不开 原因是用的C盘用户文件夹里的jupyter 而不是D盘里的 删掉后就正常了
但是比较迷 jupyter notebook的主题和图标啥的都还有 不知道是在哪里保留的 按理说C盘文件删了应该没了呀
哎 算了 先记录一下安装流程 谁知道后面还要重装几次 这台电脑已经装了n次Anaconda了
记录
pip install torch==1.5.0+cpu -f https://download.pytorch.org/whl/torch_stable.html
pip install tensorflow==1.12 tqdm numpy==1.16.4 pandas==0.24.2 scikit-learn requests matplotlib pytest h5py
pip install h5py==2.8.0rc1
创建虚拟环境
tf1.4
conda create -n tf1.4 python=3.6.5
activate tf1.4
pip install tensorflow==1.4 tqdm numpy==1.16.4 pandas==0.24.2 scikit-learn requests matplotlib pytest h5py==2.10.0
tf1.15
conda create -n tf1.15 python=3.6.5
activate tf1.15
pip install tensorflow==1.15 tqdm numpy==1.16.4 pandas==0.24.2 scikit-learn requests matplotlib pytest h5py==2.10.0
tf2.1
conda create -n tf2.1 python=3.6.5
activate tf2.1
pip install tensorflow==2.1 tqdm numpy==1.16.4 pandas==0.24.2 scikit-learn requests h5py==2.10.0 matplotlib pytest
py38
conda create -n py38 python=3.8
activate py38
pip install tensorflow tqdm numpy pandas scikit-learn requests matplotlib pytest h5py==2.10.0
conda create -n tf1.4 python=3.6.5
pip install tensorflow==1.4 tqdm numpy==1.16.4 pandas==0.24.2 scikit-learn requests matplotlib pytest h5py==2.10.0
conda create -n tf1.13 python=3.6.5
pip install tensorflow==1.13.1 tqdm numpy==1.16.4 pandas==0.24.2 scikit-learn requests matplotlib pytest h5py==2.10.0
conda create -n tf1.14 python=3.6.5
pip install tensorflow==1.14 tqdm numpy==1.16.4 pandas==0.24.2 scikit-learn requests matplotlib pytest h5py==2.10.0
conda create -n tf1.15 python=3.6.5
pip install tensorflow==1.15 tqdm numpy==1.16.4 pandas==0.24.2 scikit-learn requests matplotlib pytest h5py==2.10.0
conda create -n tf2.1 python=3.6.5
pip install tensorflow==2.1 tqdm numpy==1.16.4 pandas==0.24.2 scikit-learn requests h5py==2.10.0 matplotlib