This repo is to practice all the code from a book, Deep Learning with Python
I highly recommend those deep learning learner to buy this book, that explains the deep learning algorithm in the simplest wording and in the most practical way to display it.
Moreover, it is using keras that can use Tensorflow, CNTK or Theano as the backend, a high level deep learning library, to run those deep learning model. Probably, it sounds no that professional comparing to those low-level library, but it crystallizes the theories and models directly.
- Download Miniconda3
wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda.sh
chmod +x Miniconda3-latest-Linux-x86_64.sh
./Miniconda3-latest-Linux-x86_64.sh
export PATH=~/miniconda/bin:$PATH- Download Nvidia Driver
sudo apt install nvidia-driver-version-number- Install CUDA Toolkit
cd ~/Downloads
chmod +x cuda_file.run
sudo ./cuda_file.run
nano ~/.bashrc
export PATH=/usr/local/cuda-version/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-version/lib64
source ~/.bashrccheck whether CUDA installs successfully
nvcc -VIt means CUDA installed successfully if it shows:
nvcc: NVIDIA (R) Cuda compiler driver- Install cuDNN
Download all cuDNN Runtime Library, cuDNN Developer Library and cuDNN Code Samples and User Guide
# example
sudo dpkg -i libcudnn7.4.2.24-1+cuda10.0_amd64.deb
sudo dpkg -i libcudnn7-dev_7.4.2.24-1+cuda10.0_amd64.deb
sudo dpkg -i libcudnn7-doc_7.4.2.24-1+cuda10.0_amd64.deb- Install Bazel if it is needed
# example
sudo apt-get install git pkg-config zip g++ zlib1g-dev unzip python
chmod +x bazel-0.15.0-installer-linux-x86_64.sh
./bazel-0.15.0-installer-darwin-x86_64.sh --usercreate conda environment
# example
conda create -n tf-gpu python=3.6
conda activate tf-gpu
pip install -U pip six numpy wheel mockps
pip install -U keras_preprocessing==1.0.3 -- no-deps
pip install -U keras_applications==1.0.5 -- no-depsDownload the source code of tensorflow
# example
git clone https://github.com/tensorflow/tensorflow.git
cd tensorflow
#switch to r1.12
git checkout branch r1.12
./configure
# Do you wish to build TensorFlow with CUDA support? [y/N]: Y
# Please specify the CUDA SDK version you want to use. [Leave empty to default to CUDA 9.0]: 10.0
# Please specify the cuDNN version you want to use. [Leave empty to default to cuDNN 7.0]: 7.4
# Please specify the NCCL version you want to use. If NCLL 2.2 is not installed, then you can use version 1.3 that can be fetched automatically but it may have worse performance with multiple GPUs. [Default is 2.2]: 1.3
#Compiled the code
bazel build --config=opt --config=cuda //tensorflow/tools/pip_package:build_pip_package
#Build the a pip package from the tensorflow build package
./bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg
#pip install the tensorflow
pip install /tmp/tensorflow_pkg/tensorflow-1.12.0-cp36-cp36m-linux_x86_64.whlTest whether tensorflow GPU supporting version is installed successfully
import tensorflow as tf
# Creates a graph.
# GPU
with tf.device('/gpu:0'):
a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b')
c = tf.matmul(a, b)
# Creates a session with log_device_placement set to True.
sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))
# Runs the op.
print(sess.run(c))
### print result:
# [[ 22. 28.]
# [ 49. 64.]]
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent