Tips for Hyperas

The project is to discover the similarity of two Chinese sentences. I use LSTM to build up the model and want to tune the hyper parameters. Of course, there are a few posts and examples about it. But I have a few unique challenges:

2. Define Function Data().

Here are some tips:

A. Please import packages in functions. Otherwise it will tell you that the package or function from the package is not defined.

B. Process the data in the data function. There are supposed to be no any codes before data function.

C. Implement clean strings or special symbols out of the file. My texts have Chinese characters and special symbols like “”?、！。. These symbols do not exist in English. So if you clean the strings in data functions with Hyperas, it would give you the error “SyntaxError: EOL while scanning string literal”. My solution is to use another file to preprocess those texts.

So I wrote the cleaned data out and imported them into the file with Hyperas. I tried to call functions but it failed. Maybe you guys can find other easier ways to deal with that.

3. Define your model. Please assign inputs in this model function. You can change any hyperparameters you want with {{choice([X,X,X])}}. Concatenate() does not affect Hyperas at all. It would be same as Sequential().

Related posts:

32When Mary came where Jesus was and saw him, she knelt at his feet and said to him, “Lord, if you had been here, my brother would not have died.” 33When Jesus saw her weeping, and the Jews who came…

Umbert loves to go to the restaurant. He is a gourmet, and every month, he gives you his choices for the best places. You can check Umbert Weekends in Paris. To behave correctly at the restaurant is…

Another decade comes to an end and it has been a great one, maybe for the human race but not really for mother nature. So why not start the new year and decade with a reality check. Nature is…