import mlflow
import pandas as pd

model = mlflow.keras.load_model("my_model")
data = pd.read_json('my_model/input_example.json', orient='split')

print(model.predict(data))