Skip to content

專案環境建置

Yi-Lin Tung edited this page Jan 18, 2019 · 6 revisions

作業系統:Ubuntu 16.04.5 LTS (Xenial Xerus) Desktop
硬體需求:NVIDIA顯示卡

  1. 設定Ubuntu Linux使用NVIDIA驅動程式
    Additional Drivers
    Using NVIDIA binary driver - version 384.130 from nvidia-384
  2. 安裝CUDA
    參考TensorFlow GPU安裝說明:https://www.tensorflow.org/install/gpu
  3. 安裝Anaconda3 Python 3.6
    https://www.anaconda.com/download/#linux
    bash Anaconda3-5.2.0-Linux-x86_64.sh
    當出現"Do you wish the installer to prepend the Anaconda3 install location to PATH in your /home/yilintung/.bashrc ? [yes|no]"請回答"no"
  4. 撰寫Shell Script(檔名:Anaconda3.sh)以設定Anaconda Python環境
    export LANG='UTC-8'
    export LC_ALL='en_US.UTF-8'
    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/extras/CUPTI/lib64
    export PATH=/home/yilintung/anaconda3/bin:$PATH
    . /home/yilintung/anaconda3/etc/profile.d/conda.sh
  5. 執行Shell Script,進入Anaconda Python環境
    source ./Anaconda3.sh
  6. 建立專案的Python 3.6虛擬環境
    conda create --name autokeras python=3.6 anaconda
  7. 進入專案的Python 3.6虛擬環境
    conda activate autokeras
  8. 安裝AutoKeras套件
    pip install autokeras

至此,專案環境建置完成;若要執行專案程式碼,請參照以下命令:
python AgentNeural.py

Clone this wiki locally
You can’t perform that action at this time.