Jupyter notebook 建立 Tensorflow(2.3.0)

# Jupyter notebook 建立 Tensorflow(2.3.0) :::success #### 目前安裝成功率 13/13 ::: # 1. 安裝Anaconda 到Anaconda網站下載anaconda https://www.anaconda.com/products/individual ![](https://i.imgur.com/hypVoF7.png) # 2. 建立conda環境打開Anaconda prompt ![](https://i.imgur.com/05ovl7o.png) - 裝環境 ```shell= conda create –n tf2.3 python=3.8 ``` - 查看環境 ```shell= conda env list ``` - 啟動環境 ```shell= conda activate tf2.3 ``` - 安裝tensorflow-gpu (ref: https://lufor129.medium.com/%E5%82%BB%E7%93%9C%E5%BC%8Ftensorflow-keras%E5%AE%89%E8%A3%9D%E6%95%99%E5%AD%B8-730b235275d) 1. GPU Driver Version ```shell= ## windows cmd nvidia-smi ``` ![](https://i.imgur.com/uPkr4E4.png) 2. 查看tensorflow-gpu版本之CUDA cuDNN的版本限制 ![](https://i.imgur.com/4VYIRIQ.png) by https://www.tensorflow.org/install/source_windows 3. 比對自己的CUDA有沒有辦法支援 ![](https://i.imgur.com/o4QFTcv.png) by https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html/ 4. install ```shell= pip install tensorflow-gpu==2.3.0 conda install cudatoolkit=10.1 conda install cudnn=7.6.5 ``` - 查看已安裝套件 ```shell= conda list ``` - 測試裝好的tensorflow 若沒有跑出任何error訊息，即tensorflow安裝正確! ```python= python import tensorflow ``` - 確認tensorflow與GPU有接好有顯示GPU device才算是有接好喔! 否則還會是用CPU在執行 <p.s>若需要downgrade用以下指令 - pip install protobuf==3.20.1 ```python= print(tensorflow.config.list_physical_devices('GPU')) #from tensorflow.python.client import device_lib #print(device_lib.list_local_devices()) ``` ![](https://i.imgur.com/GvtHnLb.png)  - 上述兩個檢測做完，即可退出python ```shell= // 退出python ctrl+z ``` - 註冊jupyter ```shell= conda install jupyter conda install ipykernel python -m ipykernel install --user --name tf2.3 --display-name "tf2.3" ``` --- - 打開jupyter notebook ```shell= jupyter notebook ``` - 安裝其他常用套件 ```shell= pip install opencv-python conda install matplotlib conda install pandas==1.2.4 seaborn==0.11.1 pip install -U scikit-learn ``` - 安裝tf.keras繪制模型圖套件 1. pydot ```python= pip install pydot ``` 2. Graphviz 1. https://www.graphviz.org/download/ ![](https://i.imgur.com/eAdeigd.png) 2. 環境變數 (假設裝在C槽底下) 環境變數加入path: "C:\Graphviz\bin"，並且restart kernel即可。