常用命令记录
2021-04-19
2 min read
训练 MNIST 模型
python mnist_saved_model.py --work_dir=/home/hadoop/workspace/dataset/mnist --training_iteration=10000 --model_version=20210419 /home/hadoop/workspace/tensorflow-serving/models
/home/hadoop/workspace/dataset/mnist 数据集所在目录
10000 迭代次数
20210419 生成的模型版本
/home/hadoop/workspace/tensorflow-serving/models 模型保存路径
启动 TensorFlow Serving 服务
sudo docker run -p 8500:8500 --mount type=bind,source=/home/hadoop/workspace/tensorflow-serving/models,target=/models/mnist -e MODEL_NAME=mnist -t tensorflow/serving
8500:8500 主机端口:容器端口映射
type=bind,source=/home/hadoop/workspace/tensorflow-serving/models,target=/models/mnist 将主机 /home/hadoop/workspace/tensorflow-serving/models 目录挂在到容器 /models/mnist
MODEL_NAME=mnist 设置环境变量 MODEL_NAME, tensorflow/serving 镜像依赖该变量查找模型文件
tensorflow/serving 目标镜像
测试 TensorFlow Serving 服务
python mnist_client.py --num_tests=1000 --server=0.0.0.0:8500 --work_dir=/home/hadoop/workspace/dataset/mnist
1000 测试的样本数
0.0.0.0:8500 TensorFlow Serving 的地址
home/hadoop/workspace/dataset/mnist 测试样本所在地址
