Conda安装不同的虚拟环境

1. 安装TensorFlow2.x

1.1 conda创建环境

conda create -n tensorflow2 python=3.9
conda activate tensorflow2

# 删除环境
conda remove -n tensorflow2 --all

1.2 安装package

# 前提：安装好CUDA11.2和cuDNN8.1
pip install tensorflow-gpu==2.5

# 安装常用包
conda install tqdm
conda install numpy
conda install pandas
conda install matplotlib
conda install scikit-learn

pip install swifter

1.3 判断GPU是否可用

import tensorflow as tf

print(tf.config.list_physical_devices('GPU'))

2. 安装DeepCTR环境

2.1 conda创建环境

Now deepctr is available for python 2.7and 3.5, 3.6, 3.7.

conda create -n deepctr python=3.7
conda activate deepctr 

# 删除环境
conda remove -n deepctr --all

1.2 安装package

# 前提：安装好CUDA11.2和cuDNN8.1
# tensorflow版本过高可能报错
pip install tensorflow-gpu==2.5
pip install deepctr[gpu]

# 若已安装了TensorFlow，则使用命令
pip install -U deepctr

# 安装常用包
conda install tqdm
conda install numpy
conda install pandas
conda install matplotlib
conda install scikit-learn

pip install swifter