TensorFlow 2.0 加载已训练模型进行测试demo
import tensorflow as tf
tf.compat.v1.enable_eager_execution()
savedModelPath = 'hdfs://hadoop_dirs/test/model_dump/disk/training_models/'
importedModel = tf.saved_model.load(savedModelPath)
# Check its architecture
print(importedModel.signatures['serving_default'].structured_outputs)
print(importedModel.signatures['serving_default'].inputs)
predict = importedModel.signatures['serving_default']
input_tensor = {
'col1': tf.constant(['test']),
'col2': tf.constant([200])
}
res = predict(input_tensor)
print(res)
本文标题:TensorFlow 2.0 加载已训练模型进行测试demo
本文链接:https://www.haomeiwen.com/subject/iobpsltx.html
网友评论